AALTO YLIOPISTO FACULTY OF PHYSICS AND APPLIED MATHEMATICS SYSTEMS ANALYSIS LABORATORY
Emission estimation of marine traffic using
vessel characteristics and AIS-data Master‟s Thesis
Lasse Johansson
9/19/2011
Supervisor: Prof. Harri Ehtamo
Instructors: Ph.D. Jukka-Pekka Jalkanen, prof. Jaakko Kukkonen
i
ii
AALTO UNIVERSITY
11000, 00076 Aalto
http://www.aalto.fi
ABSTRACT OF THE MASTER‟S
THESIS
AUTHOR: Lasse Johansson
TITLE: Emission estimation of marine traffic using vessel characteristics and AIS-data
TITLE IN FINNISH:
Laivapäästöjen mallinnus AIS-dataan ja aluskohtaisiin tietoihin perustuen
FACULTY: Faculty of information and natural sciences
DEGREE PROGRAMME: The department of physics and applied mathematics
MAJOR:
Systems analysis (F3008)
MINOR
Industrial management and engineering
Supervisor: Prof. Harri Ehtamo
Instructors: Ph.D. Jukka-Pekka Jalkanen, Prof. Jaakko Kukkonen
Based on the STEAM model at the Finnish Meteorological Institute, an extended method is
presented for the evaluation of the exhaust emissions of marine traffic. The model uses
detailed technical data of each individual vessel and also messages provided by the Automatic
Identification System (AIS), which enable the positioning of ship emissions with a high spatial
resolution (typically a few meters). The presented Ship Traffic Emissions Assessment Model
(STEAM2) allows for the influences of accurate travel routes and ship speed, engine load,
fuel sulfur content, multiengine setups, abatement methods and waves.
The previously developed model was applicable for evaluating the emissions of NOx, SOx and
CO2. The extended version addresses also the mass-based emissions of particulate matter
(PM) and carbon monoxide (CO). Compared to the previous version, a more detailed power
and fuel consumption estimation process is introduced and the model‟s performance is
evaluated against available experimental data on engine power, fuel consumption and the
composition-resolved emissions of PM.
Several ways for improving the model further are also presented. This includes the addition
of kinetic energy of the individual ships, the addition of sea currents and a sub program to
process AIS data before it is used for modeling.
Finally, the model is used for estimating emissions and shipping statistics from marine traffic
in the Baltic Sea during 2006 and 2009 and based on these results, the most contributing flag
states and ship types are identified. The geographical shifts in the marine traffic are visualized
with a difference map and the effect of a sulfur reducing legislation is quantified.
Date: 19.9.2011 Language: English Pages: 81
Keywords: emission estimation, STEAM, AIS, marine traffic, particle matter emissions
iii
AALTO YLIOPISTO
11000, 00076 Aalto
http://www.aalto.fi
DIPLOMITYÖN TIIVISTELMÄ
TEKIJÄ: Lasse Johansson
TYÖN NIMI: Laivapäästöjen mallinnus AIS-dataan ja aluskohtaisiin tietoihin perustuen
TIEDEKUNTA: informaatio- ja luonnontieteiden tiedekunta
OPINTO-OHJELMA: Teknillinen fysiikka ja matematiikka
PÄÄAINE: Systeemianalyysi (F3008)
SIVUAINE: Teollisuustalous
Valvoja: Prof. Harri Ehtamo
Ohjaajat: Ph.D. Jukka-Pekka Jalkanen, Prof. Jaakko Kukkonen
Tässä työssä esitellään laajennettu ja paranneltu versio laivojen pakokaasupäästöjen
mallintamiseen perustuen Ilmatieteen laitoksen laivapäästömalliin (STEAM). Suurille laivoille
pakollisen automaattisen viestijärjestelmän (AIS) ja kattavan laivatietokannan avulla laivojen
paikka- ja nopeustiedoista voidaan laivoille estimoida päämoottorien kuormitusasteet ja
polttoaineen kulutus. Ottaen huomioon mm. käytetyn polttoaineen laadun, mahdolliset
asennetut päästövähennyslaitteistot ja laivaa ympäröivän aallokon, voidaan mallilla estimoida
laivojen pakokaasupäästöjen maantieteellistä jakautumista ja kokonaispäästömääriä Itämerellä.
Mallin aikaisemmin julkaistu versio sisälsi hiilidioksidipäästöjen lisäksi ihmisille haitallisten
rikin ja typen oksidien mallintamisen. Työssä esiteltävä laajennettu päästömalli estimoi näiden
lisäksi myös laivojen partikkeli- ja hiilimonoksidipäästöt. Lisäksi, laajennetussa mallissa
käytetään paranneltua tehon ja polttoaineen kulutuksen estimointitapaa, joka ottaa huomioon
laivan fyysiset mitat. Laajennetun mallin ennustustarkkuutta verrataan kokeellisesti mitattuja
teholukemia vastaan. Tämän lisäksi laivayhtiöiden antamia vuosittaisia polttoaineen
kulutusarvoja verrataan mallin tuottamiin ennusteisiin.
Mallin tuottamia Itämeren laivapäästötuloksia vuosille 2006 - 2009 esitellään työssä
yksityiskohtaisesti. Vuosina 2008 ja 2009 Euroopan taloutta ravisutelleen laskusuhdanteen
vaikutukset laivaliikenteeseen arvioidaan sekä Itämeren maat järjestetään niiden laivastojen
päästömäärien mukaiseen järjestykseen. Lisäksi, laivaliikenteen maantieteellisiä muutoksia
tutkitulla aikavälillä havainnollistetaan sekä vuonna 2006 voimaan tulleen merkittävän
rikkidirektiivin vaikutuksia esitellään.
Työssä esitetään lukuisia tapoja joilla laajennettua mallia voidaan jatkokehittää, kuten
merkittävien merivirtojen sisällyttämisen malliin, AIS-datan esikäsittelyn sekä laivojen
kiihdytysvaiheen tarkemman mallinnuksen.
Päivämäärä: 19.9.2011 Kieli: englanti Sivumäärä: 81
Avainsanat: emission estimation, STEAM, AIS, marine traffic, particle matter emissions
iv
v
TABLE OF CONTENTS Emission estimation of marine traffic in the Baltic Sea ...................................................
Master‟s Thesis .................................................................................................................
Abbreviations ........................................................................................................... viii
Symbols and notations ............................................................................................... ix
Acknowledgements ..................................................................................................... 0
1. Introduction ................................................................................................................ 1
2. Exhaust emissions of diesel engines and their effect on human health .................... 4
2.1.1 Carbon monoxide (CO) and carbon dioxide (CO2) ........................................ 4
2.1.2 Nitrogen oxides (NOX) ...................................................................................... 5
2.1.3 Sulfur oxides (SOX) ............................................................................................ 5
2.1.4 Particulate matter (PM) ..................................................................................... 6
2.2 Abatement methods and reduction potential of emissions ..................................... 6
3. Extended emission estimation model (STEAM2) .................................................... 8
3.1 Input data for the model .......................................................................................... 8
3.2 Power and ship attribute estimation ....................................................................... 10
3.2.1 Resistance and friction from moving in water ................................................ 13
3.2.2 The evaluation of auxiliary power ................................................................... 15
3.3 Engine load and the specific fuel oil consumption (SFOC).................................. 16
3.4 Multi-engine installations ........................................................................................ 18
3.5 Exhaust emissions modelling ................................................................................. 19
3.5.1 NOX emission factor ........................................................................................ 19
3.5.2 PM emission factor .......................................................................................... 20
3.5.3 SOX emission factor ......................................................................................... 24
3.5.4 CO2 and CO emissions modeling ................................................................... 25
4. Further improvements for the extended model version ......................................... 27
4.1 Currents ................................................................................................................... 27
4.1.1 Implementing the effect of sea currents to the model ................................... 30
vi
4.2 Kinetic energy and acceleration ............................................................................. 30
4.2.1 Engine power estimates with acceleration near harbor area .......................... 32
4.3 Acceleration based component for carbon monoxide emission estimation ........ 33
4.4 The quality issues of AIS data ................................................................................ 35
4.4.1 AIS data processing ......................................................................................... 35
4.4.2 Inactive unidentified ships ............................................................................... 37
4.4.3 AIS coverage correction .................................................................................. 39
4.5 The revised relation between PM-emissions and SFOC ...................................... 40
4.5.1 Emissions coefficients based on emissions per consumed fuel ..................... 41
4.5.2 SOX emissions using the new PM emission coefficients ................................ 43
4.6 NOX emission modelling using combustion time .......................................... 44
4.6.1 Reaction time for NOX formation ................................................................... 44
4.6.2 Temperature in NOX formation process ........................................................ 46
4.7 Alternative marine fuel – Liquid Natural Gas (LNG) ........................................... 47
4.7.1 Emission modeling with ships using LNG fuel .............................................. 48
4.8 Route interpolation with a shortest path algorithm ............................................... 50
5. Model evaluation ...................................................................................................... 53
5.1 Evaluation of the predictions of STEAM2 for engine power ............................... 53
5.2 Evaluation of STEAM2 predictions for fuel consumption ................................... 55
5.3 Evaluation of the modeling of load balancing in STEAM2 .................................. 56
5.4 Evaluation of the PM emission factors .................................................................. 58
6. Emission analysis for the Baltic Sea shipping .......................................................... 61
6.1 Emissions from Baltic Sea shipping in 2006-2009 ................................................ 61
6.2 Geographical emission changes from 2006 to 2009 .............................................. 65
6.2.1 Allocation of emission costs by geographical area ......................................... 67
6.3 Flag state analysis .................................................................................................... 67
6.3.1 Allocation of emission costs by region or flag state ........................................ 69
6.4 Emission analysis by ship type and size ................................................................. 70
vii
6.4.1 Allocation of emission costs by individual fuel consumption ........................ 73
7. Conclusions ............................................................................................................... 74
7.1 Conclusions from the emission estimation model ................................................ 74
7.2 Conclusions from the emission estimates for 2006 - 2009 ................................... 75
7.3 Further improvements ............................................................................................ 77
References ......................................................................................................................... 79
viii
Abbreviations
Organic carbon
Elemental carbon
Nitrogen oxides
Sulphur oxides
Carbon monoxide
Sulphate
Carbon dioxide
Particle matter, defined as the sum of OC,
EC, Ash, and its associated water
STEAM Ship Traffic Emission Assessment Model
RPM Revolutions per minute
AIS Automatic identification system
Helcom Helsinki Commission, Baltic environmental
protection commission
IMO International Marine Organization
IMO GHG2 IMO Greenhouse gas study, published in
2009
SECA Sulfur emission control area
TEU Twenty-foot equivalent unit, a measure used
for capacity in container transportation
SFOC Specific fuel-oil Consumption
CAT Caterpillar, engine manufacturer
ix
RoPax Roll-on, Roll-off passenger ship
RoRo Roll-on, Roll-off vehicle carrier ship
MMSI Maritime mobile service identity, a series of
nine digits which are sent in digital form over
a radio
GT Gross tonnage of a ship
DWT Deadweight of a ship
LOA Length overall in meters
LBP Length between perpendiculars in meters
LNG Liquid natural gas
Symbols and notations
Resistance force resulting from moving in
water (turbulence, waves, displacement of
water)
Propelling force of the ship
Resistance force component which is in
parallel to the speed vector. is caused by
the surrounding environment (wind, air and
ambient waves)
Resistance force resulting from friction
induced to the wet surface of the ship
Quasi-propulsive constant, a ship specific
efficiency ratio of the ship‟s propeller which is
the ratio of and generated shaft force of the
engine
x
( ) Speed of the ship
( ) Acceleration
( ) Effective acceleration of the ship which is
caused by engine use
( ) Instantaneous power requirement in kW
Service power rating of a ship
Number of revolution per minute of a
propeller
Propeller diameter
Block coefficient
( ) Froude number
Engine load
Base value for the specific fuel-oil
consumption of an engine
Relative value of SFOC compared against the
minimum value
Mass of a ship, defined as the sum of gross
tonnage and deadweight
Mass of the particle in grams
Conversion rate of the particle in
combustion process, particles formed per
time interval
Emission factor measured in g/kWh for CO
Base value for carbon monoxide emission
factor
xi
Emission factor measured in g/kWh for total
PM emissions
Revised emission factor for PM emissions
Acceleration based component of
emission factor
Velocity vector of a current
Speed component of which is in parallel
with the ship‟s velocity
Time, combustion time
Stroke type dependent coefficient for
determining combustion process time
Ratio of heat values of LNG and diesel fuel
oil
The mass of produced by combusting
one gram of LNG fuel
The mass of produced by combusting
one gram of diesel fuel
Acknowledgements
I would like to thank the Finnish Meteorological Institute for giving me this opportunity
to contribute to their novel way of modeling ship emissions. I am grateful for the open
minded and helpful atmosphere of the co-workers (especially Jukka-Pekka Jalkanen,
the original creator of STEAM model) which made possible to edit the model source
code quite freely to gain better insight of the model itself. I would also like to thank
Jaakko Kukkonen for his editorial contribution and for the guidance which has made
me a better writer in scientific literature. Furthermore, it would not have been possible
to produce such an extensive representation of the model along with its validation data
by the author himself and it should be noted that especially chapters 3 and 5 contain a
significant contribution from the instructors of this thesis.
1
1. Introduction
Emissions from shipping have a significant impact on ambient air quality in densely
populated coastal areas. Indeed, emission smog rising from the busiest ship routes in
the world are clearly visible from space. The marine exhaust emissions may
substantially contribute to detrimental impacts on human health; a study made by
Corbett et al. estimates that the annual premature mortality caused by shipping will be
approximately 87 000 premature deaths annually in 2012 if shipping emissions are not
further regulated (Corbett et al., 2009). A more recent study, however, suggests that the
contribution of shipping might be even greater - Particle matter ( ) emissions alone
are estimated to be responsible for the loss of 1.8 million years of life lost including 1.3
million years of life lost due to mortality in the studied six countries of Belgium,
Finland, France, Germany, Italy and the Netherlands, making the most significant
environmental factor affecting public health (Hänninen et al. 2011). The contribution of
shipping sector to the amount of in coastal urban areas such as Helsinki,
Rotterdam and Los Angeles have been estimated to be significant, ranging from 13% up
to 42% depending on the geographical area of measurements (Starcrest, 2008).
Besides particle emissions, marine shipping causes significant carbon monoxide ( ),
nitrogen and sulfate oxide emissions ( , ), which in turn affect the public health
and increase the amount of acid rain. Moreover, sulfur and nitrogen oxides in marine
exhaust gas have been reported to affect cloud formation process in the lower
atmosphere due to the so-called indirect aerosol effect – the particles are able to act as
cloud condensation nuclei (Schreier et al., 2006).
The yearly monetary cost of the health problems that are associated with marine traffic
is a subject of debate in recent literature. Depending on the contribution of marine
traffic to the total emissions, the yearly monetary cost is likely to be in several
billion of euros in Europe alone. Because of the high costs, stringent limits for the
sulphur content of marine fuels and for emissions are being introduced, which are
expected to reduce the emissions from ships. As a result, emissions would be
simultaneously reduced, as a major part of emissions is in the form of sulphate.
However, sulphur content reductions will not eradicate emissions completely
(Winnes and Fridell, 2010; Fridell et al., 2008; Cooper, 2006; Cooper, 2003; Kasper et
al., 2007; Buhaug et al., 2009), even if the global fleet would switch to low sulphur fuel.
2
The emissions of can also be reduced by using after-treatment techniques, which
will remove a significant part of the emissions (Corbett et al., 2010; European
Commission Directorate General Environment, 2005) Scrubbing systems from engine
manufacturers have been commonly applied to diesel power plants on land, but their
commercial installations to ships have been scarce. This is expected to change, after the
implementation of the stringent sulphur limits included in the revised Marpol Annex
VI of the IMO (International Maritime Organization, 1998).
Policy makers cannot effectively reduce health problems caused by marine traffic
without sufficient and timely information about the total amounts and geographical
distribution of the emissions. Therefore emissions are estimated with several different
models that produces information about where, how much and by whom emissions are
being released into the atmosphere. Currently available global ship emission inventories
are mostly based on top to down (i.e., top-down) –approaches in which emission
estimates are made using fuel statistics and known ship routes without considering single
vessel characteristics. However, the statistics concerning the sales of marine fuels are
difficult to disaggregate to the amounts of fuel burned regionally or locally. The
approaches based on fleet activities, called as bottom-up methods, have therefore
recently gained popularity; new ship emission inventories have been generated
especially for arctic regions (Paxian et al., 2010; Corbett et al., 2010). Various regional
ship emission inventories have been introduced (Matthias et al., 2010; De Meyer et al.,
2008) and the previously significant uncertainties in the estimated emissions of global
ship traffic have been evaluated to have decreased during the last half decade (Paxian et
al., 2010; Lack et al., 2008). Information is currently scarce especially regarding the
geographical distribution and chemical composition of emissions arising from ship
traffic, and the chemical composition details have not commonly been introduced to
global bottom-up inventories of ship emissions.
In 2009, presented at the Finnish Meteorological Institute, a bottom-up method for the
evaluation of the exhaust emissions of marine traffic was introduced. The model was
based on the messages provided by the Automatic Identification System (AIS), which
enable the identification and location determination of ships (Jalkanen et al., 2009).
The use of the AIS data facilitates the positioning of ship emissions with a high spatial
resolution, which is limited only by the inaccuracies of the Global Positioning System
(typically a few meters). Using the AIS data for emission estimation almost totally
3
removes the inaccuracies that are caused by the uncertainties of evaluating the times of
ships spent at sea and at berth. The instantaneous speeds of the vessels are also known
from the AIS data, the use of which substantially reduces the uncertainties in analyzing
the operational states of the ship engines. The methodologies for evaluating the power
and fuel consumption, however, were fairly simple in the introduced model version,
and the assumptions of the model were observed to provide biased estimates especially
for auxiliary engines. In this paper, an extended version of the emission estimation
model is presented.
The main objectives of this thesis are (i) to introduce and evaluate the extended model
and its new features for marine emission modeling, (ii) to present ways to improve the
extended model even further and (iii) to present emission estimations and shipping
traffic analysis produced by the model about marine traffic in the Baltic Sea. The
modelled emission types and their effect on human health are briefly discussed in
Chapter 2. In Chapter 3 the extended model with its principles and mathematical
structure, including the model‟s recent adjustments such as more sophisticated scheme
for the resistance evaluation and a load balancing of the engine, are introduced.
Chapter 4 includes the discussion of the challenges that have been encountered after
extensive use of the extended model and also, some solutions for these problems and
new modifications to the model are presented. Most of the changes for the model that
are suggested in this chapter however, are currently being implemented and thus are not
affecting emission estimates presented in this thesis.
The main results of this thesis are presented in Chapter 5 and 6, followed by discussion
and conclusions of the results in Chapter 7. In chapter 5, the model‟s capability to
predict instantaneous power requirements and fuel consumption is evaluated against
available experimental data and emission factors provided by the model are compared
to the available measurements presented in recent literature. In Chapter 6 the total
emission estimates for the Baltic Sea shipping between January 2006 and December
2009 are presented. As a byproduct to the emission estimate outputs, the statistical data
produced by the model is put into use and the flag states and ship types, that are making
the biggest contribution, are identified. Also, geographical shifts in marine traffic during
the recent years are identified and the effect of the late recession in Europe is studied.
4
2. Exhaust emissions of diesel engines and their effect on
human health
In the diesel engine combustion process, high pressured fine droplets of diesel fuel are
mixed with air and the mixture spontaneously combusts after being heavily pressurized..
For a typical 2-stroke marine diesel engine, which uses 170 grams of fuel per produced
kWh, 7.8 kg of air is used as the combustion process requires large amount of oxygen
(21% in volume of air). In the combustion process, volatile carbon compounds react
with oxygen forming carbon dioxide and water, simultaneously releasing significant
amount of heat. The exhaust gas (170g of fuel per kWh) contains approximately 0.5 kg
of , 0.2kg of vaporized water and also 1.1kg of excess oxygen (Kuiken, 2008). The
increase in heat in the combustion chamber causes thermodynamic expansion and
piston movement, which is converted to mechanical work for the rotating crankshaft.
The movement of the pistons causes friction and thus a constant supply of lube oil is
needed. For a typical engine, the need of lube is a couple of grams per kWh but may
contain heavy and alkaline metals, which are released in the exhaust gas as the lube oil
gets perfectly combusted in the process as well.
Besides and the small amount of heavy metals, significant amount of harmful
material is released to the atmosphere in the exhaust gas that causes direct and indirect
detrimental effects to human health and to the environment, such as , , ,
and . These pollutants and their effects on human health and to the environment
are briefly discussed here. The pollutants react in atmosphere with various complex
chemical interactions. Some chemical interactions start immediately in the early plume
phase making the measurement of emissions directly from exhaust plumes challenging.
These chemical reactions and their effect thereafter, however, are beyond the scope of
this thesis.
2.1.1 Carbon monoxide (CO) and carbon dioxide (CO2)
is the main product resulting from the combustion of fossile fuel and is not harmful
for human health but is well known for its effect on increasing the radiative forcing on
Earth‟s surface [ ]. In other words, enforces the phenomenon called the
greenhouse effect.
5
Carbon monoxide is a colorless and odorless gas which is a product of incomplete
combustion of organic matter due to insufficient oxygen supply to enable complete
oxidation to carbon dioxide. Mild exposures to may cause headaches, vertigo, and
flu-like effects, but larger exposures (at 100ppm or greater) can lead to significant
toxicity of the central nervous system and heart, and even death. (Prockop L,
Chichkova, R., 2007).
2.1.2 Nitrogen oxides (NOX)
Air is mostly (78%) nitrogen and thus large amount of nitrogen gets mixed in the
combustion process. Nitrogen is an inert gas that doesn‟t under normal circumstances
combust with other substances. In diesel engine combustion process, however, the
temperature rises up to 2000 , which is more than sufficient to enable nitrogen
combustion.
Nitrogen oxides are precursor components for a photochemical reaction in which
ozone is formed in the lower atmosphere (Brunekreef et al., 2002). Nitrogen dioxide is
also a catalyst compound for the formation and thus for acid rain. Exposure to
ozone can lead to a variety of respiratory health effects, such as coughing, throat
irritation and reduced lung function. In addition, it can worsen bronchitis, emphysema,
and asthma. (Hänninen et al., 2011)
2.1.3 Sulfur oxides (SOX)
Currently in 2011 inside the sulfur emission control area (SECA) which contains the
Baltic Sea, marine diesel-oil contains sulfur compounds up to 1% of mass and before
May 2006 the maximum allowed sulfur level was as much as 2.7%. The combustion
process generates significant amount of sulfur oxides, especially and . When
released in to the atmosphere, is further oxidized, usually in the presence of a
catalyst such as , forming and thus is causing acid rain. Emitted
particles with its direct and indirect effects are making the largest non-greenhouse
contribution of shipping to global climate forcing. (Petzold, 2010). The direct aerosol
forcing of sulfate particles from shipping ranges between and
and thus balances the positive radiative forcing effect of .
The range of the released particles is limited to a few hundred kilometers
depending on weather and wind conditions. Because of this, most of the emissions
6
at open sea end up being neutralized by sea water and are not believed to pose a direct
threat to human health.
2.1.4 Particulate matter (PM)
Unlike other emissions, which are chemically defined, particulate matter ( ) is
defined in international standards (ISO 8178) as the mass that is collected on a filter
under specified conditions. The chemical composition as well as the size distribution,
however, varies in literature. In this paper, (diameter ) from marine
emissions is defined to include ash particles, organic ( ) and elementary carbon ( )
and sulphate ( ) and its associated water molecules. Fuel sulphur content has been
observed to affect significantly to the amount of emissions because all of the
sulphate and its associated water molecules originate from the sulphur content of the
fuel.
is the most thoroughly internationally reviewed environmental pollutant during the
last decade. The health implications of the particulate matter components have been
extensively studied and convincing epidemiological evidence associates mass
concentrations with the health impacts (Hänninen et al. 2011). Exposure to has
been associated with both respiratory and cardiovascular effects and total mortality.
emissions have been estimated to cause a loss of 1.8 million years of life lost annually,
including 1.3 million years of life lost due to mortality in the studied six countries of
Belgium, Finland, France, Germany, Italy and the Netherlands (Hänninen et al. 2011).
Overall 67 % of the estimated environmental burden of disease in the study of
Hänninen et al. was explained by exposure to making it the most significant
environmental factor affecting public health.
2.2 Abatement methods and reduction potential of emissions
Although it is possible to remove from exhaust gases by chemical conversion, this
is not considered feasible. A number of options for improvements in efficiency and
thus for reducing fuel consumption and however, are readily available. The
potential for fuel savings can be very significant (20% - 75%), but at the same time the
costs, lack of incentives and other barriers may prevent many of the efficiency
improvements from being adopted (Buhaug, 2009).
7
Significant reductions of emissions can be achieved through limitations on the
sulfur content of fuel. Other possibility to reduce the level of is the installation of
an exhaust-gas scrubbing system. Two main principles exist for the scrubbing system:
open-loop seawater scrubbers and closed-loop scrubbers. Both scrubber concepts may
also remove and limited amounts of (Buhaug, 2009). Emissions that are
removed from the exhaust are carried in the wash water and released to the sea. After
being released, sulphur oxides react with the seawater to form stable compounds that
are normally abundant in seawater and are not believed to pose any danger to the
environment and human health.
emissions and especially sulphate particles can be reduced by scrubbing with
seawater. Claims for the potential reduction of emission levels range from 90% to
20% depending on the particle size distribution as the smaller emission particles are
very difficult to extract from the exhaust gas. Emissions of can be further reduced
by optimizing combustion process, for example by achieving better oxidization and
minimizing of the lube oil consumption as well as the additives in lube oil itself. The
burning of fuel–water emulsions can also reduce emissions of to a certain extent,
but may cause reductions to the efficiency of the engine.
Emissions of from diesel engines can be reduced by a number of measures. One
possibility is the reduction of combustion temperature with fuel modification, e.g., water
emulsion or with the humidification of the charge air. Other options include exhaust
gas recirculation (EGR), the modification of the combustion process and its timing and
also the treatment of the exhaust gas with selective catalytic reduction (SCR). However,
the purity of water is an issue with all options that use water and the sulfur content
influences many of the possibilities for emission-reduction technologies including EGR
and SCR (Buhaug, 2009).
8
3. Extended emission estimation model (STEAM2)
In this chapter, the extended model, Ship Traffic Emission Assessment Model
(STEAM2), is presented and especially the new features since the previous model are
illustrated in greater detail.
Previously at the Finnish Meteorological Institute a method was presented for the
evaluation of the exhaust emissions of marine traffic, based on the messages provided
by the Automatic Identification System (AIS), which enable the identification, location
and speed determination of ships (Jalkanen et al., 2009). The model used an internal
ship database that included, among other specifications, engine power, maximum speed
and weight and spatial attributes. The model was based on the relationship of the
instantaneous speed to the design speed and the use of the detailed technical
information of the engines. The effect of waves was also included in the model. The
previously developed model was applicable for evaluating only the emissions of ,
and .
The main differences between the new model (STEAM2) and the previously
developed one (STEAM) include that the - and emissions are included in the
latter model. A new evaluation method is also used for analyzing the resistance of ships
in water. The model also includes a refined modeling of the power consumption of
auxiliary engines, which depend on ship type and its operation mode and also the effect
of engine load to fuel consumption. An illustration of the main components of the
STEAM2 model is presented in Figure 1. The main input data sources are the internal
ship database and the AIS data. Based on the properties of the ships and its power
requirements, the model can evaluate the power consumption and load of the engine,
and then the fuel consumption of the ship. Based on these values, the model is used to
evaluate the emissions of , , , and , as a function of time and
location.
3.1 Input data for the model
The automatic identification system (AIS) was introduced by the International Maritime
Organization to enhance safety and efficiency of navigation, safety of life at sea and also
for maritime environmental protection through better identification of vessels
(Mokhtari 2007). The main objective, however, was to provide precise information that
could be used to collision avoidance. AIS equipment operates on VHF frequency and
9
is able to detect other AIS transmitters nearby. AIS message includes static information
about the identification (IMO and MMSI-number), name and ship type, dynamic
information such as the position, heading and speed over ground and also voyage
related information about the type of cargo and destination.
Figure 1: A schematic diagram of the main components of the STEAM2 model and their inter-relations. The model input data sources are presented on the uppermost row
of rectangles, and the model output data (i.e., emissions) are presented on the lowest row of rectangles. The arrows describe either the flow of information in the model, or a modeled dependency between various factors. The different colors denote the various categories of factors included in the model; dotted and solid arrows are used only for
visual clarity.
Often some of the data fields of lesser importance to the collision avoidance purposes,
such as cargo type, are left blank but nevertheless the use of the AIS data facilitates an
accurate mapping of the ship traffic, including the detailed instantaneous location and
speed of each vessel in the considered area. For example, more than 210 million
position reports were received from the 9497 AIS targets in the Baltic Sea in 2007. For
ships in a regular schedule, this results in tens of thousands of position updates each
month. Decrypted AIS data, usually with time interval between two consequent
messages of no less than 3 minutes, is provided by Helcom for the model. Currently for
10
a typical month, the amount of AIS message inputs for the model amounts to more
than 2 gigabytes of data and therefore, even when much more frequent AIS data would
be readily available, messages with longer time interval are favored because of
computational limitations.
The internal ship database of the STEAM2 model contains the technical details of
ships used in the evaluation of emissions. The database contains the information of
more than 30 000 ships; this is approximately a third of the global fleet. Most of the
ships in the database are newer ships that have been built within the last two decades;
most of these ships are frequently operating in the Baltic Sea. If ships with new IMO
signature are appearing in the AIS messages, then a query is sent to HIS Fairplay and if
the particular ship has been listed there, then the ship with its specifications from IHS
Fairplay is added to the internal ship database before the emission estimation program
is run. Especially for older ships it would be wise to update ship specifications from
time to time, but due to high costs updating isn‟t done regularly. In case a ship sending
AIS messages cannot be identified using the ship database and with queries to IHS
Fairplay, then the ship is assumed to be a small tugboat vessel and the model uses
generic average attributes of small tugboats for the unidentified ship.
3.2 Power and ship attribute estimation
A method presented by Hollenbach (1998) is used to calculate the resistance of ships
due to moving in water. The predictions of the Hollenbach method agree well with
other performance prediction methods, such as those of Holtrop-Mennen (Matulja and
Dejhalla, 2007; Holtrop and Mennen, 1982; Holtrop and Mennen, 1978). The use of
this method, compared with the previous model, improves the predictions of resistance
and engine power, especially in cases, in which the hull dimensions and the engine data
is available, but the design speed of the vessel is unknown. In the previous version of
the STEAM model, the design speed was a critical parameter for the model
performance; if that value was not available, an average speed was used instead that was
specific for each ship type. The use of the Hollenbach method avoids such
assumptions, and therefore provides a more reliable basis for the resistance
calculations.
11
Figure 2: A marine vessel moving in water and the forces acting on it. is the
propelling force of the engine, is the resistance from moving in water including
the forming of waves. is the friction between the wet surface of the ship and the sea.
is the resistance force component in parallel to which is caused by wind, sea ice
coverage and ambient large waves.
In Figure 2, a marine vessel moving in water with constant speed and the forces acting
on it are presented. Waves tend to increase the fuel consumption of ships no matter
what direction the waves are propagating in relation to the ship. The model takes the
effects of waves into account by using the hourly significant wave heights and the mean
wave direction data from a separate wave model. Besides the sea state, the Force
induced by waves depends on parameters describing the wet surface and the three-
dimensional structure of the hull and also of the contact angle between the hull and
waves (Jalkanen, 2009). Currently, is set to include only the resistance from waves
because air resistance in most cases have been calculated to be miniscule compared to
other forces and moving through ice coverage is yet to be implemented.
The vessel of Figure 2 is in equilibrium state as the acceleration is zero and therefore
the sum of forces is also zero:
(1)
Where is the propelling force of the engine, is the resistance from moving in
water, is the resistance from ambient large waves which is in parallel to and
𝐹𝐹𝑟
𝐹𝑊𝑎𝑡𝑒𝑟 𝐹𝐸𝑛𝑣
𝐹𝑃
12
is the friction of the water surface. Taking into account that ( )
( )
( ) the propelling force can be written as
( )
( )
( )
(2)
where ( ) is the current engine power, ( ) is the vessel speed and is the quasi
propulsive constant. The dimensionless quasi propulsive constant is used to describe
the effectiveness of converting the main engine power to actual propelling power, taking
propulsive losses arising from transmission, hull, shaft and propeller itself into account.
Finally, multiplying Eq. 1 by ( ) yields
( )
( )( )
(3)
According to empirical studies of Watson (1998), can be estimated indirectly from
the ship attributes in the following manner:
√
(4)
where is the rpm of the propeller and is the length between
perpendiculars. The physical dependencies between efficiency, propeller rpm and
are quite complicated and are best left outside of this thesis. Propeller efficiency is
commonly substantially less than unity; usually 60-80 % of the main engine power is
transmitted to the water by the propeller (Watson, 1998).
As Eq. 4 shows, a value for propeller rpm is required to estimate the propeller and
transmission losses and the required main engine power. If the number of propellers is
unknown, then the ship is simply assumed to operate with a single propeller. Propeller
diameter is estimated using the method described by Watson (1998). An estimate for
propeller diameter in meters is
(5)
where is the service power of the main engine (80 % of the maximum continuous
rating provided by IHS Fairplay (2010) in kilowatts and is the propeller‟s rpm. This
13
method was used for all single-propeller vessels, for which the propeller rpm was
known. For multi-propeller vessels, if both the propeller rpm and diameter were
unknown, a value was used that is based on a fraction of vessel draught. This approach
does not consider exceptional cases of surface piercing propellers. It is expected to lead
to a reasonable estimate of propeller diameter. In multi-propeller cases and also if
propeller data is unavailable, propeller size is estimated with a ship type specific fraction
of draught, as draught is one of the main limiting factors for propeller size. These
fractions of draught values have been statistically estimated using the internal ship
database.
If propeller rpm cannot be determined from ship technical data and it cannot be
estimated using Eq. 4 and 5, the power is predicted based on the previous version of
the model (Jalkanen et al., 2009). In these cases, 80 % of the main engine power is
assumed to be in use, when the vessel is traveling at its design speed. The required
power is computed applying a relationship , where is a ship-specific constant
generated from main particulars and is the instantaneous speed of the vessel.
In the internal ship database sufficient propeller details exist for about 60 % of the
cases, which facilitate the evaluation of the quasi propulsive constant. In the remaining
cases, the previous method (Jalkanen et al., 2009) of engine power estimation for the
main engines has to be used, which requires that the design speed of the ship has to be
known. In approximately five percent of the ship database entries both the propeller
rpm and vessel design speed are missing. In such cases, the emission predictions are
relatively less accurate, as average values specific to this ship type have to be used as a
substitute for the missing ship data values. The values larger than the total installed
engine power are not allowed for by the model.
3.2.1 Resistance and friction from moving in water
Modern ship designing and building has its roots dating back to the Middle Ages and
because of this, even today many empirical rules of thumb involving lots of different
spatial parameters and coefficients are being followed by ship builders as a first design
phase. This is reflected on the Hollenbach method as well, which is used to calculate
and based on the physical dimensions of the ship. The method itself is
based on the resistance measurements of 433 tank tests. The application of the method,
however, is in many cases limited by the availability of the hull and propeller details and
14
for reliable power predictions using this model, either propeller revolutions per minute
(rpm) or propeller size has to be known.
The resistance from moving in water is mainly affected by a coefficient called as the
block coefficient which reflects a resistance factor originating from the shape of the
hull. Roughly put, more hydrodynamic hull has a smaller than a bulkier one. The
block coefficient cannot be obtained from commercial available databases however, and
thus is needed to be estimated with a method suggested by Watson and Gilfillan
(1976) and further described by Townsin (1979): The Block coefficient can be written
as
(
) , (6)
where is the Froude number. The Froude number is defined as the ratio of a
characteristic velocity to a gravitational wave velocity or equivalently as the ratio of a
body's inertia to gravitational forces. It is used to determine the resistance of an
object moving through water and is computed in the Hollenbach method as follows:
( )
√
(7)
Sadly, the waterline length in Eq. 7 isn‟t readily available for most of the vessels. In
these cases, an approximation using the average value of overall length in meters ( )
and length between perpendiculars in meters ( ) is used instead. Thus, The Froude
number is given by:
( )
( )
√ ( )
(8)
The residual resistance that is approximately the sum of and is given by:
( )
( )
(9)
where is the density of seawater and is a ship specific factor, which is also speed
dependent as the displacement of water caused by the ship is dependent on speed and
is the wet surface of the ship. The Hollenbach method essentially estimates using
hull and ship specific parameters, the block coefficient and wet surface . Using Eq.
15
4 and Substituting given by the Hollenbach method to Eq. 3 yields the following
formula for power estimation:
( )
( )( ( ) ( ) )
√
(10)
where ( ) is a dimensionless function of the Froude number and is a ship specific
constant which is affected by the ship‟s dimensions, number of bossings and hubs, etc.
3.2.2 The evaluation of auxiliary power
Just like in the previous version of the model, the main principle of auxiliary power
estimation is a classification by ship type and its operation modes. There are three
different operation modes available for each ship: maneuvering near harbor areas,
cruising and hoteling. The operation mode is currently being deduced from the ship‟s
speed data. The different ship types are discussed in greater detail in Chapter 6.
In the previous model version, especially the estimation of auxiliary power was
observed to be insufficient, and thus the following modifications were made: Passenger
class vessels which includes cruise ships, Roll-in, Roll-off vehicle carriers (RoRo),
vehicle passenger ships (RoPax) and yacht vessels use a base value of 750 kW of
auxiliary engine power for all operating modes, but an additional requirement of 3 kW
is added for each cabin. This emulates the additional need for electricity required by air
conditioning, hot water and other electrical installations inside the cabins. For reefers
and containerships, similar assumptions are applied. A base value of 750 kW is used
while cruising, 1000 kW during hotelling and 1250 kW while maneuvering. In addition
to these values, each refrigerated Twenty-foot Equivalent Unit (TEU, standardized
cargo container) consumes approximately 4 kW of electricity to maintain the containers
in a constant temperature. Clearly, the actual power requirement of the container
depends on the temperature difference between the environment and the container
(Wild, 2009).
All other vessel classes use 750, 1000 and 1250 kW for cruising, hotelling and
maneuvering, respectively. With these modifications, STEAM2 can distinguish between
large and small vessels of the same ship type. However, in all cases, the installed
auxiliary engine power is used as an upper limit for the predicted auxiliary engine power
(in cases, for which the computed auxiliary power would exceed the installed auxiliary
16
power). Boiler energy usage is included in the estimates of auxiliary engine power; these
have not been modeled explicitly due to the lack of data.
3.3 Engine load and the specific fuel oil consumption (SFOC)
Instantaneous total fuel consumption is influenced by many independent factors. Fuel
consumption of main engines used in propulsion is commonly estimated in available
literature as a product of the constant specific fuel oil consumption (SFOC) and
instantaneous engine power, which results in a linear relationship between fuel
consumption and engine power. Ideally, all power systems that require fuel to operate
should be modeled separately, such as the main engines for propulsion, the auxiliary
engines for power generation and the boilers for heat generation. However, in practice a
separate modeling of all of these is currently not feasible.
Figure 3: The relative specific fuel-oil consumption (SFOC) as a function of the relative engine load, based on the data of three engine manufacturers: Wärtsilä, Caterpillar and
MAN. The data of Caterpillar is based on three different SFOC curves of small 4-stroke engines, and the data of MAN is based on large 2-stroke engines.
The relative SFOC curve provided by the engine manufacturer Wärtsilä for a medium
sized 4-stroke engine is presented in Figure 3. Using SFOC studies and engine
specifications (Caterpillar, 2010; Man B&W, 2010), two other relative SFOC curves by
other manufacturers are also presented. The engines by MAN considered here are
17
large 2-stroke models, whereas the Caterpillar engines are relatively small 4-stroke
models.
For all three curves presented, the SFOC is a non-linear function of engine load, and
this function has a minimum at a specific engine load. For the data of Caterpillar, MAN
and Wärtsilä, the minimum is approximately at the relative engine load of 70, 75 and
80%, respectively. Minimizing fuel oil consumption therefore requires engine loads
approximately from 70 to 80 %. There is an approximately parabolic dependency
between the SFOC and the engine load.
In the STEAM2 model, a parabolic SFOC function for all engines is assumed. Using
regression analysis of the comprehensive SFOC measurement data from Wärtsilä, a
second degree polynomial equation for the relative SFOC was derived:
( ) (11)
where is the engine load ranging from 0 to 1. The absolute fuel consumption is
estimated from
( ) ( ) (12)
where is the so-called base value for SFOC which is a constant for each
engine. According to second IMO greenhouse gas report (Buhaug et al., 2009), a lower
consumption is assigned for new engines, describing the technical development and
better efficiency of modern engines. The base value is also influenced by engine stroke
type and power. Primarily, the engine-model specific base values for SFOC from the
engine manufacturers are used but in case such a value is not available, the value is
evaluated (taking the above mentioned factors into account) according to the IMO
GHG2 report (Buhaug et al., 2009). Depending on the stroke type, age and power
rating of the engine, the base SFOC value typically ranges between 170g/kWh and
220g/kWh.
For simplicity, it has been assumed that engine load and SFOC dependence from
Equation 11 applies to all engines. For turbine machinery, of 260 g/kWh is
used. Auxiliary engine was set to 220 g/kWh and the same load dependency
was applied.
18
In some ships, a diesel-electric engine setup has been installed, in which a diesel engine
is used to generate electricity for auxiliary engines and also for electric motors to drive
the propellers without any mechanical contact. The main engine is usually operating
constantly on an optimal engine load state, which can be regarded as one of the most
notable advantages of diesel-electric setups over traditional marine diesel engines. Thus,
the optimal SFOC value is assumed for every diesel-electric setup in the model.
3.4 Multi-engine installations
While it is straightforward to estimate an engine load of a single engine ship, if required
power is known, this estimation is more challenging for multi-engine setups, which are
common especially among passenger ships. Therefore, a load balancing scheme for
multi-engine installations has also been implemented in the STEAM2 model.
Load balancing is a crucial issue for the proper functioning of multi-engine installations.
Engines that are not needed at a specific moment can be turned off, which saves fuel
and ensures that the remaining engines are operated with an optimal engine load. To
simulate this operation of the engines, the STEAM2 model determines the minimum
number of engines, which need to be in operation to overcome the predicted resistance
of the ship. The model assumes all main engines to be identical, a minimum number of
engines are assumed to be used, and the load values are assumed to be less or equal
than 85 %. The latter assumption is needed, as engine loads larger than 85% are
commonly avoided. If this load values would be exceeded, an additional engine is
assumed to be taken online and the load is balanced among the operational engines.
For example, let us consider a ship with four installed engines, each with a power of
6000 kW, and an instantaneous power requirement of 11000 kW. The minimum
requirement to obtain 11000 kW would require operation of two engines at 91.7% load
level, which is not feasible. The modeling assumption is therefore that three engines
would be used instead, each with a load of 61.1%.
A limitation of this approach is that the model treats all main engines as equal and
neglects engine setups, for which one engine in a pair is larger than another. For
instance, in case of four engines with two pairs of identical engines, a so-called 2+2
setup, the accuracy of the predictions of fuel consumption and emissions will
deteriorate. Passenger classed ships are required to have at least two engines operational
at all times, due to chief engineer interviews and vessel safety rules. Load balancing is
19
applied to both main and auxiliary engines, but in case of diesel-electric engine setups,
all the power commonly required for ship systems and propulsion are taken from the
main engines.
3.5 Exhaust emissions modelling
Using the estimates for current instantaneous power, fuel consumption and ship
attributes taken from ship database, emissions of every emission type presented in
Chapter 2 are estimated for ship routes derived from the AIS data. Ship routes are
generated from the sequence of coordinates simply using a linear interpolation between
consecutive coordinates for each ship with unique identification signature. The main
aim is that the model provides accurate emission factors for the all pollutants, including
all the chemical components of , for all values of the fuel sulphur content and the
engine load. The evaluation of the influence of engine load is needed especially for an
accurate description of emissions of , , and . All emissions have
therefore been assumed to be dependent on engine load, except for those of .
3.5.1 NOX emission factor
Besides high temperature, the amount of combusted nitrogen is strongly dependent on
time allowed for combustion reaction, and thus is affected directly by the RPM value of
the engine. Therefore, high speed engines produce less emissions per kWh than
low speed engines as the duration of high temperature phase is shorter. In several
studies it has been concluded that emissions are not strongly affected by fuel
consumption and thus are left independent of SFOC in the model.
emissions are calculated in the model using a fixed emission factor [g/kWh],
which is determined by the main engine‟s rated speed (RPM) value using the
emission factor curve from IMO (Tier I), which reflects the current limits for the
allowed emission factors as a function of engine‟s rated speed. In the forthcoming
years, however, marine vessels are expected to meet even lower emission factor limits
(given by Tier II and Tier III curves) than the model assumes. These three emission
factor limits as a function of RPM are presented in Figure 4.
As the three emission curves in Figure 4 illustrates, emissions tend to
increase significantly in the lower RPM regime. A typical 4-stroke engine has an RPM
20
of more than 500 revolutions per minute while it is not uncommon that a powerful 2-
stroke engine is running slower than 100 revolutions per minute.
Figure 4: Emission Limits set by MARPOL Annex VI. The upper curve (Tier I),
since January 2000, applies retroactively to all engines with power more than 150kW. Curve in the middle (Tier II), applies after January 2011. The third curve (Tier III),
which applies only for SECA areas, comes into effect on January 2016.
Starting from January 2011, Tier II standards are expected to be met by combustion
process optimization for each new ship. Most of the engine manufacturers, however,
have been producing Tier II compliant engines in anticipation of the Tier II limitations
for the last decade. It is not likely that an engine has significantly lower emission
factor without having some abatement method installed, which in that case would be
specified in the ship database and taken into account. Moreover, for each new ship, the
engine‟s emission factor is certified (with ISO-8178 test) to agree with the IMO
Tier II curve. Thus, the maximum allowed value for emission factor given by Tier
II curve can be applied for new ships with reasonably good accuracy, although with a
slight possible overestimation.
3.5.2 PM emission factor
As discussed previously, fuel consumption is dependent on engine load and thus the
emissions of several pollutants have the same dependency. If the engine is run outside
its normal operating load range, fuel consumption and thus also emissions are
increased, since the engines are not commonly optimized to run on low loads for
21
prolonged periods. In case of multi-engine setups, however, unnecessary engines can be
turned off to gain better efficiency in fuel consumption as it was discussed in Chapter
3.3.
Several authors have reported experimental results on the composition of particulate
matter as a function of engine load (Agrawal et al., 2008; Agrawal et al., 2010; Petzold et
al., 2008; Moldanova et al., 2009; Sarvi et al., 2008) and sulphur content (Sarvi et al.,
2008; Buhaug et al., 2009) and this relationship between emissions, engine load
and sulphur content has been implemented into the model.
Figure 5: The emission factor of the total PM, and for its chemical constituents as a function of fuel sulphur content (mass-based percentage), based on the data from the
second IMO GHG study (Buhaug et al., 2009). Linear regression curves are presented as black lines. The emission factors of the total PM, SO4 and H2O are linearly
dependent on the fuel sulphur content. The data points for EC and ash are partly overlapping in the figure.
The sulphur content of the fuel has a crucial influence on the emissions. The
dependency of emission factor on fuel sulphur content was modeled according to
Buhaug et al. (2009), as presented in Figure 5. As expected, the emission factors of the
total , and its associated water molecules are linearly dependent on the fuel
sulphur content, whereas the emission factors of , and ash are almost totally
independent of this factor. The emissions of could therefore not be eradicated
22
totally, even if sulphur would be completely eliminated from ship fuels (Winnes and
Fridell, 2010; Buhaug et al., 2009).
Applying linear regression analysis to the data presented in Figure 5 yields the following
emission factor dependencies:
and
( )
( )
g/kWh
(13a)
(13b)
where is the fuel sulphur content in mass percentages and the emission coefficients
for , and ash have been assumed to be independent of the sulphur content.
However, the amount of ash may change between different fuel grades. Comparing the
atomic masses of and its associated water in Eq. 13a implies that each
molecule is associated with approximately 4.2 molecules of according to the IMO
GHG study. The total emission factor (in g/kWh) is assumed to be the sum of the
above mentioned emission factors:
( ) ( ) (14)
In STEAM2, the emissions [g/kWh] are evaluated as the product of specific fuel-
oil consumption and emission factors. For example, the total emission as a function
of engine load and fuel sulphur content is
( )
( ) ( )
(15)
where the relative is computed using Eq. 11. The variations of this emission
factor have been graphically illustrated in Figure 6. In short, the variation of the
emission factor for different components has been modeled based on the variation of
SFOC. The emissions of all components are modelled based on the variations of
SFOC and instantaneous power, and in addition the emission factors of sulphate and
associated water are dependent also on the fuel sulphur content.
23
Figure 6: The predictions of the STEAM2 model for total PM emission factor [legend, in units of g/kWh] as a function of engine load and fuel sulphur content.
It was concluded that the behavior of one of the constituents behave differently
with very low engine loads. In two separate studies made by Agrawal (Agrawal, 2008)
and Petzold (Petzold, 2008) emission factor for was observed to increase up to
0.6g/kWh with very low engine loads. Such an increase was not explained through the
increase in SFOC as the other components in the studies were not affected by low
engine loads nearly as much. Thus, it was concluded that emissions are depended
on engine load with very low engine loads. Based on these measurements using
regression analysis, as a special rule, the emission factor for was set to equal
{
(16)
emissions are effectively multiplied with low engine loads through the combination
of SFOC increase and Eq. 16, although it should be highly uncommon for any large
ship to be sailing with such a low engine load for economic reasons.
24
3.5.3 SOX emission factor
emissions originate from the residual sulphur of the combusted fuel and thus the
emission factor for depends mainly on the current fuel consumption and therefore
is affected by engine load as presented in Chapter 3.3.
All of the sulphur molecules in fuel is assumed to be converted either to sulphate ( )
or sulphur oxides ( , ). The amount of is calculated the conservation of
matter –principle: The amount of sulphur in grams for any consumed fuel amount
must satisfy
(17)
where coefficients and are the relative sulfur-mass content of the particles. One
sulphur atom weights . The ratio of and particles is assumed to be one
and thus the average mass of is . Using the atomic mass of 96 for sulphate,
the coefficient is calculated to equal 0.444 and is equal to 0.333. For any fuel
consumption [g/kWh] the amount of sulphur per kWh is
and thus, combining
Eq. 17 and 13a yields
(18)
The amount of sulphate depends on and SFOC as well and thus the ratio of ( )
and is constant. It can be calculated that for any given SFOC the model uses the
following ratio:
(19)
Taking into account the atomic weights of and Eq. 19 can be presented in an
equivalent form using the conversion ratios, yielding the following conversion rate for
sulphate:
25
(
)(
)
( )
( )
(20)
where
is the number of converted particles per time unit and the conversion rate
is assumed to be constant. According to the model then, for every sulphate molecule
there is approximately 18 particles formed simultaneously.
3.5.4 CO2 and CO emissions modeling
Assuming perfect combustion conditions, the amount of emitted can be estimated
in a straightforward manner from the amount of fuel burned. However, the
emissions are substantially more dependent on engine load. The data based on three
experimental studies and the modeled dependency of the base emission factor of as
a function of engine load has been presented in Figure 7. The base emission factor
as described by Sarvi et al. has been adopted in STEAM2.
Figure 7: The base value of emission as a function of relative engine load. The
measurements of Agrawal, Moldanova and Sarvi have been shown, and the base
emission factor curve is based on Sarvi. The emissions of are also influenced by
rapid changes of relative engine load.
During normal engine operation, when engine load ranges from 75% to full load, the
base emission factor of is below 0.75 g/kWh according to Sarvi (2008). However,
using the engine at low engine loads will significantly increase the emission factor.
26
Moreover, a rapid change of engine load has been observed (Cooper, 2003; Cooper,
2001) to result in significantly increased emissions of carbon monoxide. This is usually
the case, when the ship is accelerating or actively decelerating (braking). Therefore, the
modelled curve (as presented above) has been modified with an additional scaling term,
which amplifies the emission factor, if the ship is accelerating.
Using this scaling factor called Acceleration Based Component (ABC), the
emissions takes the following form:
(21)
where
*
+
(22)
For simplicity, the empirical factor has been assumed to be the same for all ships, =
600. The ABC is 1.0, if there is no significant acceleration; otherwise it is larger than
unity. Strictly speaking the ABC value should be ship-dependent. More experimental
data would be needed to model these relationships in more detail. Moreover, the
modeling above cannot distinguish between natural deceleration (engines stopped) and
active braking (ship using its engines to decelerate). Therefore, the emissions might
be over-predicted in case of natural deceleration. In the following chapter, the ABC
factor for emissions is discussed in greater detail.
27
4. Further improvements for the extended model version
In order to estimate emissions correctly and associate them with correct geographical
areas, (i) ships are needed to be identified and associated with correct specifications, (ii)
the instantaneous power is needed to be estimated correctly and (iii), the emission
formation processes are needed to be modeled correctly using the derived fuel
consumption and engine state. It is not surprising that such a model is susceptible to
host of different sources of error as none of the three procedures i-iii is easy to execute
with high precision.
Even though the prediction accuracy of the model has been observed to be satisfactory
(Chapter 5), it can be made better and more consistent. Therefore the model and its
features have been under re-evaluation. Sources of error have been identified and also
totally new features are planned to be implemented as several of these have been
estimated to have a significant impact on emission estimates and prediction error.
These potential sources of error, shortcomings and new possible features are discussed
in this chapter.
4.1 Currents
The Baltic Sea has regular currents and streams throughout the year, even though the
strength of these currents may not be comparable with those of the Atlantic or North
Sea. The speed of these water flows, however, is large enough to affect the power
estimates significantly. The whereabouts of the most notable currents in the Baltic Sea
are presented In Figure 8.
In order to take the effect of the current into account, the speed of the vessel in relation
to flowing water is needed to be known. For a ship sailing in a massive water flow in
Figure 8, the velocity of the vessel v can be estimated to be the sum of current‟s velocity
and the ship‟s velocity relative to the water :
(23)
If the speed of the current compared to the ship‟s speed is small, or if is small,
then and are almost in parallel and
(24)
28
Figure 8: Major sea currents in the Baltic Sea. The sea currents and their direction are presented by red arrows. In the picture, a marine vessel with a GPS-measured speed of
is moving in current. is the speed vector of the current and is the angle between
and .The map (back picture) is provided by the European Environment Agency.
where is the parallel speed component of in relation to v (and approximately to
v‟ as well). the effective strength of the current is the scalar projection of onto given
by:
( )
(25)
where is the angle between the ship‟s and the current‟s velocities. The angle can be
calculated using the heading information ( ) of the ship given by the AIS-
29
message. Thus, the speed caused by engine use can be calculated with the speed and
heading information from the AIS data for any given
Most of the notable currents at the Baltic Sea flow with a speed of no more than 0.25
meters per second, while the strongest currents in between Oland and Bornholm are
flowing with a speed around 0.5 m/s heading towards Denmark. In Chapter 3.2, it was
shown that the current engine power is strongly affected by the speed: where
is some ship specific constant. Therefore, the relative effect of the addition of currents
to the model estimates is
(26)
Using Eq. 26 reveals, for example, that if a ship with a typical speed of 9 m/s is moving
in parallel against a strong current of 0.5 m/s, then the power estimate would be
increased by 17.6% by the addition of currents. The currents in other sea regions, for
example, at the North Sea are much stronger and the speed of water flow may rise up
to 1 m/s. For a ship with a speed of 9 m/s sailing head-on against such a strong current,
the power consumption could be up to 37% more due to currents than is currently
estimated. Indeed, in an unpublished study, for a large RoPax vessel sailing from
Rotterdam to Harwich in September 2009, the model was seen to produce significantly
biased estimates while the ship was sailing near known currents. In this study, while the
engine load was kept constant the speed of the ship gradually increased for more than
1.5 m/s.
The major ship routes in the Baltic Sea seem to coincide with the most notable
currents, which can be seen comparing Figure 8 and Figure 20a-b in Chapter 6. Indeed,
strong currents are systematically taken advantage of by seasoned navigators. Thus it is
likely that significant corrections to estimated power might be applied frequently by the
addition of currents and also that the net effect on the power estimates would be
negative in terms of average power. With other sea regions, however, the possibilities
for taking advantage of the currents are not so simple. For example, avoiding currents
of the English Channel by selecting another route is not always economically feasible.
30
4.1.1 Implementing the effect of sea currents to the model
As it was shown, it becomes increasingly important to take possible currents into
account if the model is to be used in other, more turbulent sea regions such as the
North Sea. The implementation would be fairly simple – All that is needed is a grid
map of the sea region that contains the information about the speed and the direction
of the current.
In the power estimation phase, the strength of currents at the location of the ship is
evaluated each time. In case a strong current exists, the effective strength of the current
is then calculated using Eq. 25. Then, the power is calculated using a corrected
speed estimate instead of .
With other sea regions, however, the implementation of sea currents might be more
complicated as the flow of the water masses is affected by tides and changes in
temperature and thus is showing a temporal dependency. Because of this, at least
separate current maps for each season might be needed.
4.2 Kinetic energy and acceleration
Fuel consumption has been underestimated with the model especially near harbor
areas and one known reason for this is the lack of kinetic energy modeling. The harbor
areas are naturally the most important areas for the emission estimation as these are the
emissions that cause most of the health problems to the general population.
Not only does the increase in kinetic energy require substantial increase in
instantaneous power output for any massive marine vessel, also the ships tend to
decelerate near the destination port using engine power to be able maneuver safely.
Even though the effect of this addition on the estimated power curve profile near
harbor areas might prove to be substantial, the total effect on fuel consumption should
be nevertheless small.
Besides the force that is needed to overcome the resistance of water, the extra thrust
force that causes the ship to accelerate is equal to
( )
( )
(27)
31
where is the quasi-propulsive constant of the ship. Instantaneous extra engine
power ( ) is also equal to the increase rate of kinetic energy:
( )
(28)
where ( ) . Therefore, the extra power that is required for acceleration is
( ) ( )
( ) ( ) ( )
( ) ( )
(29)
If ( ) is negative which is the case when the ship is decelerating the ship might be
actively breaking using additional engine power. For this reason, the effective
acceleration ( ), which is the result of active engine use, needs to be defined.
Effective acceleration is simply equal to acceleration if ( ) . On the other hand, if
the ship is decelerating, then
{
( )
( )
( ) * ( ) ( )+ ( )
(30a)
(30b)
where ( ) is the total resistance from moving in water and mass is the sum of the
ship‟s gross tonnage and deadweight. Combining Eq. 30a and 30b, the effective
acceleration can be presented in the following form:
( ) { (
( )
( )) ( )
( ) ( )
(31)
It is not uncommon for large ships to accelerate up to a speed of 9 m/s being still
relatively close to harbor area. For example, if such a vessel would happen to weight a
typical amount of 60 000 tons with a value of 0.7, the acceleration would require
more than 1100kWh of energy. Indeed, during acceleration significant spikes have
been observed in power measurements taken onboard several vessels.
32
4.2.1 Engine power estimates with acceleration near harbor area
The effect of adding kinetic energy to the model was tested in principle by generating
acceleration values using speed and time data and adding ( ) to the power
estimatation phase in the model. In Figure 9 estimated power profiles with and without
( ) for a Roll on/Roll off vehicle carrier ship (RoRo) arriving at a harbor in May
2010 are presented. During 14:30 – 19:00 the ship is relatively close to human
population and the effect of acceleration seems to be notable during that time. Two
points of active braking in can be identified near 14:40 and more notably near 15:10
but the resulting power spikes are relatively small, however.
Figure 9: The predictions of the model with (green surface) and without (orange
surface) acceleration based power component. The data is from a RoRo class ship sailing near harbor area in May 2010. Smoothed speed data is presented with an orange
curve.
The speed curve presented in Figure 9 is a smoothed average:
( ) for
each time interval because using either end point value for the whole interpolation
route might result in substantial under- or overestimation of fuel consumption during
rapid speed changes. Also, the average value is needed to estimate ( ) accurately.
For example, using the current speed value in Eq. 29 for a ship which has accelerated to
any speed from a total halt during a single message interval (which happens
occasionally) would result in zero ( ).
33
At first when the addition of kinetic energy modeling was being tested, a small number
of power spikes of abnormal height were produced. Investigation revealed that the
acceleration calculations for these spikes were all based on messages with time interval
of mere seconds while the usual time interval was a couple of minutes. Most of these
messages implied impossibly rapid speed changes to have occurred during the trip.
Based on these experiments with acceleration data it was observed that filtering these
kinds of “false messages” out of the input data, if possible, would result in better input
data quality. In Chapter 4.4, the quality issues of the input data are discussed in greater
detail.
The effect of adding kinetic energy to original power estimates was also tested against
measured power consumption values. This resulted in an 4% increase in total fuel
consumption which was originally underestimated for slightly more than 9%. Still, not
all of the most significant measured power spikes near harbor areas were succesfully
predicted. It was concluded that there was some source of systematic error of greater
importance in the predictions. One of the un-modeled interactions acting on the ship
that could cause such variations in speed while the engine load was kept steady
according to direct engine load measurements taken onboard is the sea currents.
Indeed, while the most substantial estimation bias was observed the ship was sailing
against a significant sea current at the time according to a static current mapping for the
Baltic Sea.
4.3 Acceleration based component for carbon monoxide
emission estimation
Rapid engine load changes have been measured to cause extremely high emission
spikes in contrast to steady engine load emissions. To account for such notable
emission spikes, an acceleration based component to emission factor was
introduced in Chapter 3.5. The main principle of this method was to associate high
absolute values of acceleration directly with emission spikes. However, the
presented method with a constant scaling factor which is to be used with every kind of
ship may be too simplistic. Furthermore, the presented ABC factor is not able to
identify active braking and thus the ABC factor may be applied when the ship is
decelerating naturally.
34
The main assumption of the ABC factor can still be assumed to hold: emissions
seem to be approximately linearly dependent on the rate of engine load change.
Therefore the scaling factor can be assumed to be a linear function of the rate
of engine load change, given by:
( ) ( )
(32)
Rapid engine loads and emission spikes have been observed to coincide only during
acceleration and without acceleration the engine load is usually kept steady for
economic reasons. Therefore, the engine load change in Eq. 32 can be associated with
the increase in power due to acceleration. Now, let the engine load change from to
due to an increase in the instantaneous power, which is causing the ship to
accelerate. Then,
( )
( )
( )
( ) ( )
(33)
where is given by Eq. 31. According to Eq. 32 the scaling factor should be linearly
dependent on speed, acceleration and on the inverse of the acceleration duration -
Three variables instead of just the acceleration of the ship. For any ship, the revised
factor should be defined as follows:
*
+
(34)
where is a ship specific constant
and is some constant scaling
factor, which might depend on the other engine specifications and it‟s condition.
Naturally, the engine and its condition play an important role in the way how rapid
engine load changes would affect emissions, but this effect is difficult to model without
very specific information about the engine. Nevertheless, the ABC factor which in the
extended model is based on a single case-study and disregards many of the physical
factors, should at least take into account the current velocity and the time interval
between the two AIS messages. Given enough measurement data, a set of coefficients
35
could be identified and applied for different ship types and engine classes, resulting in
much more reliable basis for emission spikes estimation.
4.4 The quality issues of AIS data
During the last few years, it has become apparent that not all received AIS data is
relevant for total emissions estimation, or even valid information about moving ships.
On the contrary, AIS data seem to contain large amount of noise, for example,
unidentified ships sending messages nowhere near any ocean surface and then just a few
minutes later, the same ship might be sending the another message thousand kilometers
away from the initial coordinates. One reason for these kinds of incidents is that some
small vessels are sending AIS messages with wrongfully calibrated transmitter. Other
explanations are still under investigation.
Other concern with the emission estimation perspective is that more and more small
vessels are installing AIS transmitters. Although this offers an opportunity to model
even greater a share of the marine activity, the majority of these vessels cannot be
identified using the internal ship database, or any other commercial information source
for that matter. Therefore, preset small vessel specifications have to be used for these
ships when emissions are being estimated.
4.4.1 AIS data processing
To prevent invalid AIS messages from causing imaginary routes and therefore causing
overestimated and geographically biased estimates, the model needs to validate the
messages that are used for emission estimation.
Previously the model performed such validity checks for every data point as a final
phase for the emission estimation. Because of way the model was programmed, the
only possibility to perform this kind of check in the emission estimates phase, was to
compare spatial separation, speed and time interval to the next node. Based on the test,
emissions are estimated from initial point to the next or are discarded. However, this
test cannot be fully trusted, because it is not possible to determine the validity of a
message based on the next message; for all we know the next message might be invalid
and not the other way around. Also, usually the route interpolation process should be
performed from point A to point C if the data for the point B is evaluated to be invalid.
Even if the invalid messages were discarded correctly by the model, they are still able to
cause unnecessary calculations and files to be created during the previous phases.
36
Indeed, the interpolation and plausibility checking process is currently vulnerable and
because of this, for example, some yacht with MMSI code 0 was reported to travel
several million kilometers in each month between January and May of 2006, which is of
course an impossible feat for any marine vessel. Fortunately, later investigation proved
that these abnormal travel distances hasn‟t distorted the actual emission estimates but
rather are causing the ship‟s cumulative travel counter to build up in some cases.
Especially for the new features, the programming logic of the model posed several
difficulties: When the model was programmed the emission estimation phase was
viewed as a separate phase. For each data point, engine power was estimated and later
on the engine power estimate is applied for the next interpolated route. Because of this,
the next data point is not accessible when engine power is estimated. Hence it was
impossible to deduce, for example, if the ship is accelerating or actively using engine
power for braking. The solution for overcoming the presented challenges was simply,
that all of the input data is to be filtered from all of the invalid ones before the
processed input data is fed to the model. After appropriate filtering it is then possible,
among other things, to calculate and add acceleration to the input data. Another
solution would‟ve been to make drastic technical changes in the programming code
itself but would‟ve required a lot of programming effort.
Finally, a sub program was programmed to perform various tasks, such as:
To Sort the AIS data by MMSI code and then arrange messages by time for
each unique MMSI code.
To perform validity checks using 3 consecutive messages at a time. Based on
pair-wise time separation, spatial distance and speed -check, trash messages are
separate. In the speed check, calculated speed between the two consecutive ship
locations is compared to the maximum speed value provided by the internal
ship database.
To filter ships with only one message sent are discarded as no route can be
interpolated.
To calculate and add acceleration based on two valid data points with the same
MMSI identification.
37
To provide average speed for two consecutive valid points (Needed for the
estimation of kinetic energy and also for overall lag-reduction in the estimated
power)
The introduction of reliable input data with acceleration offers new possibilities for the
model as well as an easy way to gain control over the input data. For example, the
maneuvering and the use of thrusters in harbor areas could be identified. Previously
such maneuvering tests would‟ve been impossible to perform in the power estimation
phase.
4.4.2 Inactive unidentified ships
In Table 1, statistics about the AIS data for the recent years is presented. The number
of messages per year has been steadily increasing while the number of different ships
sending these messages is increasing even more rapidly – the number of different ships
encountered in the AIS data has more than doubled between 2006 and 2010. One
might assume that marine traffic in terms of travel amounts and fuel consumption in the
Baltic Sea has been increasing significantly as well but this is not the case as it turns out
in chapter 6.
Table 1: Annual statistics of the archived AIS messages and the annual amount of active ships. Active ship refers to a true, fuel consuming ship while all ships is the number of
total ships encountered in the annual AIS data. Active IMO ships represent the regular and certified ship traffic. For ships without IMO number, no reliable commercial
specification data has been available and are thus assumed to be small vessels.
Year Archived
messages
Temporal
AIS coverage
Active ships
(All ships )
Active IMO
ships
Active ships without
IMO number
2006 >171 966 000 93.36% 8160 (10810) 6851 (84.0%) 1309 (16.0%)
2007 >210 345 000 97.90 % 9326 (11780) 7355 (78.9%) 1971 (21.1%)
2008 >247 793 000 96.13 % 10589 (14098) 7311 (69.0%) 3278 (31.0%)
2009 >261 088 000 99.20 % 11606 (16385) 7422 (63.9%) 4184 (36.1%)
2010 >233 705 126 91.99% 12951 (22172) 7355 (56.8%) 5596 (43.2%)
In Table 1, the column “Active IMO ships” represent the number of specified, active
ships (fuel consuming according to the model) for which the use of AIS equipment and
registration with the IMO to obtain a unique registry number is mandatory. The
amount of these ships between 2006 and 2010 has increased no more than 8%.
Comparing the increase of active ships with the increase in annual amount of all
38
encountered ships reveals that the amount of singular AIS messages and unknown
vessels with inconsistent route data has become a frequent issue to deal with. The main
reason for this increase in vessels is that smaller ships are installing AIS messaging
equipment more often and most likely, sometimes with incorrect settings.
In the recent years, also the amount of active ships without proper IMO number has
increased. These vessels have to be modelled without accurate ship specifications. As it
was presented in Chapter 3.1, an unidentified ship is assumed to be a small tug boat.
The unidentified ship is attributed with a weight of 700 tons and a relatively powerful
main engine of 2300 kW, given by average specification of the listed known tug boats.
This assumption, however, results often in low engine load estimates. Indeed, recent
emissions estimations presented in Chapter 6 have shown a significant increase in
emissions for tug boats only, which (with the Sarvi‟s emission factor curve) implies
very low engine load estimations for the ship type in general.
It should be safe to assume unidentified vessels to be considered to be small ships with
an average weight of couple of hundred tons, but the average attributes of these
unknown vessels should be re-evaluated. It turns out that active ships without an IMO
number are encountered most frequently during summer which suggests that large
amount of these vessels are possibly passenger ships. However, it is difficult to obtain
any other information about these vessels. One possibility is to use the new sub-
program that processes the input data to check for each unknown vessel, for example,
how fast it travels, what kind of trips the ship makes and based on this, make more
sophisticated guesses about the attributes of the ship. Until more accurate estimates are
available, at least the main engine power should be reduced to a level which will not
cause these unidentified ships to travel with low engine load estimates.
The total contribution of tug boats and unknown vessels measured in fuel consumption
is nevertheless small (10% in 2009) and would most likely be even further reduced after
correcting the preset values for unknown vessels. Therefore, if the emission estimation
model would be expanded further and used e.g. for computationally heavy dispersion
modeling later on, it might be wise to separate these ships from the rest of the input
data altogether to reduce the amount of calculations.
39
4.4.3 AIS coverage correction
The yearly temporal coverage of AIS messages presented in Table 1 varies around 92-
99% for the past few years containing several long blackout periods, some with
identifiable causes (e.g. severed cable in a construction site in June 2006). Also,
blackouts for approximately 24h have not been uncommon, but the source of these
blackouts is still being investigated. In any case, AIS shortages are often needed to be
taken into account when total emission estimates are produced with the model.
In order to take these temporal AIS blackouts into account, it is needed to understand
the interpolation logic of the model. The interpolation logic of the model works as
follows: If the distance of subsequent messages A and B is lower than 150km and no
more than a day have passed, then STEAM2 model interpolates a travel route for the
ship from A to B if and only if the ship with its listed maximum speed could sail from A
to B in time. Temporal AIS coverage based correction of emission amount is then
rather straightforward a procedure for long blackout durations but gets more
complicated if the time interval contains a large amount of short blackouts - During
short AIS blackouts, some routes are still being interpolated resulting in a rough,
undervalued emission estimates for the time period in question.
Figure 10: Hourly number of AIS messages received between May and June of 2010. The blacouts presented in the figure are approximately 23h long and are commonly
followed by one hour of full coverage during 00:00 – 01:00.
40
In 2010 short AIS blackouts of 23 hours due to some unknown technical problem with
the servers of Helcom were so frequent that the total coverage of several months
dropped below 60%. The AIS coverage during the most notable blackouts in 2010 is
presented in Figure 10.
To correct the total emission estimates it was needed to solve how much of the ship
traffic the model was able to interpolate over the systematic blackout periods. After
several test runs it was concluded that of the daily ship traffic is interpolated
by the model for each blackout that lasts approximately one day. In the test procedure,
estimated emissions over several time intervals with full AIS coverage were compared
against respective time intervals with systematic AIS blackout sequences using the same
days of week in comparison. When emission estimates for the year 2010 was finally
produced, the importance of taking the interpolation feature into account was over 10%
of monthly emissions for some months with very low AIS coverage.
4.5 The revised relation between PM-emissions and SFOC
Engine efficiency is defined as the produced shaft power per energy in the supplied fuel
and thus the SFOC value can be viewed as a measure of efficiency of the engine.
Engines with higher base SFOC values need to consume more fuel per produced kWh
because of the lower efficiency due to engine wear, suboptimal design and combustion
timing, etc. Also, the cylinder diameter is affecting the efficiency significantly and
therefore small diesel engines have an efficiency of 25% whereas that of very large
engines exceeds 50% (Kuiken, 2008). In all situations the combustion process can be
assumed complete independent on engine load and SFOC.
In this paper, it has been presented that emissions are affected by engine load
through SFOC curve meaning essentially that increase in fuel consumption is associated
with an increase in emissions. The amount of emissions, however, is calculated
with fixed emission factors measured in g/kWh even though the amount of fuel
combusted in the process could be quite different depending on engine load and
between different engines. This raises a question – would it not be more meaningful to
measure emission coefficients as emissions per fuel burned [g/kg]? Indeed, in
some studies in literature, emission factors are presented in grams per fuel burned, but
in many cases even these coefficients are derived directly from g/kWh values using a
fixed SFOC value. Some studies even seem to deliberately avoid the discussion arising
41
from problem of variable fuel consumption per power output. This makes it difficult to
compare measured emission coefficients if the used engine‟s fuel consumption is not
specified.
The conservation of matter –principle suggests that the sulphur content of the fuel has
to be converted to either sulphur oxides or to sulphate and the amount of these
compounds must be directly linked to the amount of fuel burned. If not, then sulphur
would start cumulating inside the engine. Also, major source of ash particles in the
exhaust gas originate from the ash content of the fuel and thus ash particles in principle
should correlate strongly with fuel consumption as well. However, the formation
reactions concerning the other particles of such as and might not be
bounded by such principles.
Currently the base SFOC value is not affecting estimation of even though it is
primary factor affecting the fuel consumption; in the internal ship database there are
base SFOC values that range from 160g/kWh to 230g/kWh.
4.5.1 Emissions coefficients based on emissions per consumed fuel
To overcome the inconsistencies with emissions and their relation to fuel consumption
mentioned above, a new base SFOC dependent emission factor for was formulated
which also affects to the estimated amount of emissions as well. It should be noted
that the suggested changes are only to make the principles of emission estimation
more consistent. More measurement data for the suggested modifications are needed to
check if the changes would actually result in more accurate estimations.
While the new suggested formulation for the emission factor produces consistent
emission amounts with the amount of fuel burned. However, the new solution involves
a problem – the current emission factors provided by IMO GHG2 are based on
measurements from Germanischer Lloyd‟s institute but unfortunately, no engine
specifications are available concerning that particular study besides the fact that the test
was performed with a 2-stroke type engine. Thus the current emission factors that are
used in the model cannot be associated with a specific base SFOC. Given the base
SFOC value, however, the total emission factor of the model could be expressed in
following manner:
42
( )( ) (35)
where coefficients and can be determined given the IMO GHG2 engine
specifications or with a new similar emission factor study measurements with provided
engine specifications. Another equivalent yet simpler way of expressing the new
emission factor is in the following form
(36)
where is the currently unknown base SFOC value for the engine that
was used in determining in Chapter 3.5.2. For now, let‟s roughly assume that the
current emission factors were determined using a typical large 2-stroke engine with a
fuel consumption of 170g/kWh. The resulting new emission factors with several
different base SFOC values are presented in Table 2.
Table 2: The new PM emission factors as a function of base SFOC and fuel sulphur content (S, in units of mass percentages).
Base SFOC 170g/kWh 190g/kWh 210g/kWh 230kWh
Factors [g/kWh]
0.05 0.056 0.062 0.068
0.2 0.224 0.247 0.271
0.05 0.056 0.062 0.068
0.312 0.349 0.385 0.422
0.244 0.273 0.301 0.330
0.3 + 0.556S
0.34 + 0.62S 0.37 + 0.69S 0.41 + 0.75S
As Eq. 36 suggests, only if the base consumption of fuel is equal to the engine of
Germanischer Lloyd study, then the fuel consumption based factor is the same as
before. Furthermore, in Chapter 3.3 it was presented that in the model the base SFOC
depends on the age, stroke type and size of the engine in a manner that was suggested in
(Buhaug, 2009). Thus, the refined way of determining the emission coefficients
would indirectly cause these factors to affect emissions through base SFOC as they
already do with and . For example, the base SFOC for a four stroke main
engine of 1000kW made in 1980 would be estimated to be 230g/kWh. The resulting
emission factors would then be in fact 35% greater than the model currently
estimates.
43
As it was shown previously that currently a large portion of the total active vessels are
not specified in the ship database and thus are assumed to be small vessels equipped
with engines similar to the one in the example. Therefore, the suggested new
emission factor would most likely increase the estimated emission amounts notably.
Indeed, in a preliminary test using AIS data from May 2010, two emission simulations
(with and without the new emission factor) were compared. The test showed that
the estimates might be approximately 15% more than the model currently suggests.
4.5.2 SOX emissions using the new PM emission coefficients
In Chapter 3.5 it was shown that the amount of is calculated subtracting the
sulphur mass of from the sulphur mass of the burned fuel. emissions are not
completely in terms with the amount of fuel burned as it has been discussed in this
chapter but however is. This has led to a variable ratio between the emission
amounts of and . Indeed, it can be seen in the annual emission estimates
presented in Chapter 6 in which the ratio of and is slightly different for every
year implicating that the ratio of and emissions is dependent on the base
SFOC of the individual ships without intended physical reason.
The amount of emissions can be calculated in a more consistent way: the
conservation of mass principle dictates that for any amount of fuel burned the amount
of sulphur in and emissions must be equal to the amount of sulphur in the
burned fuel and thus, for any given fuel amount the emissions can be calculated
using the new emission factor from Table 2. Using these new, more fuel
consumption dependent emission factors, it can be calculated that the ratio between the
two emission masses (
) discussed in Chapter 3.5 would be reduced to 11.5
(and even less than 11.5 if the base SFOC of the Germanischer Lloyd study was
actually lower than 170g/kWh). Hence the conversion ratio of sulphur to sulphate
would now be slightly larger than it is currently in the model but this time, unaffected by
base SFOC as intended.
To make matters even more complicated, in a recent study by Petzold it was presented
that for several test engines the sulphate conversion ratio followed a linear, increasing
trend as a function of engine load (Petzold, 2009). In the study, sulfate emissions
increased by a factor of 3 when engine load was raised from low load (20-25%) to high
44
load (75-85%). However, the conversion rates were observed to be lower than in the
model with all engine load points. Thus it is possible that sulphate emissions and
therefore also emissions are even more dependent on engine load and are being
overestimated especially in the lower engine load states. Hopefully, more measurement
data of emissions with different load points is available in the future and the effect
of engine load to sulphate conversion can be verified and taken into account if
necessary.
4.6 NOX emission modelling using combustion time
The purpose of the proposed modification presented in Chapter 4.3 was to harmonize
the assumption how the engine load and fuel consumption together affects
emissions. In contrast to emissions, it was stated in Chapter 2, that emissions
are not affected by engine load nor fuel consumption, but depend rather on the
engine‟s rated RPM. Based on the latter, it is natural to assume that the instantaneous
emissions depend on the current RPM of the engine. However, the current RPM
is actually a function of the engine load, which implies that emissions are indeed
affected by engine load. It would seem that the assumptions of formation reactions
need to be harmonized as well, and hence the nitrogen reactions with oxygen are
studied here with greater depth.
In typical marine diesel engine combustion, approximately 170g of fuel, 1,3g of lube oil
and 7.8 kg of air is used per each produced kWh. 78% of air is nitrogen and 21% is
oxygen. The exhaust gas usually contains approximately 0.5 kg of , 0.2kg of
vaporized water and also 1.1kg of excess oxygen (Kuiken, 2008). Nitrogen reaction
requires time, oxygen and a high temperature. Clearly, the abundant oxygen cannot be
the bottleneck for nitrogen reactions and thus there are left only two factors to account
for – time and temperature.
4.6.1 Reaction time for NOX formation
According to Sarvi, for any given two engine states ( , ) the relation of engine
load and speed for a typical marine diesel engine is as follows:
(
)
(
)
(37)
45
Substituting with the maximum rated and fixing then with the value of
1, the of the engine as a function of engine load is given by
√ (38)
where and are engine specific constants, which determine the minimum
engine speed that is needed for the engine to run on low engine loads. For Eq. 38 to
give meaningful values, the following must apply for full engine load with maximum
:
( )
(39)
For the test engine used by Sarvi, the coefficients can be derived with linear regression:
a = 0.74 and = 133.6. Indeed, the engine load in the Sarvi‟s study seem to
drop to below 0.25 with speeds lower than 60%% of the rated . Studies made by
Agrawal shows that Eq. 38 used with these coefficients agrees with at least 3 other
engines of different sizes.
In a 2-stroke engine, the total combustion cycle is done once for each revolution of the
crankshaft whereas the 4-stroke engines use two revolutions for the total cycle. The total
combustion process including expansion takes approximately 120 crank degrees and
with 4-stroke engine about 140 degrees according to Kuiken (Kuiken, 2008). Thus, the
reaction time for nitrogen reaction is given by:
(40)
where is for 4-stroke and for 2-stroke. Using Eq. 38 and
40, the reaction time can be calculated for any engine with a specified speed rating.
Currently, the rated RPM value of the main engine is not specified for only 3.73% of
the ships in database.
The number of revolutions per minute varies significantly; from 100 up to 2000
revolutions and thus the reaction time for formation varies even more between
engines as the fastest engines are usually 4-stroke engines with a smaller value of . In
Figure 11, measured emission factors from four separate studies (Agrawal 2008,
46
2009, 2010), (Sarvi, 2008) are presented in function of calculated reaction time given by
Eq. 40.
Figure 11: Measured emission factors plotted against calculated reaction time.
Measurements are from 4 separate studies conducted by Agrawal (3 studies) and Sarvi (one study). Also a regression power series (black line) for the combined measurements
are presented. All the engines are Tier I type.
The correlation for the regression line is just above 0.5. Indeed, a linear regression line
would do just as well for the data presented in Figure 11 but in general the
emissions from high speed engines are proven to have a significantly lower emission
factor and therefore the regression line should decrease rapidly as reaction time
approaches zero. Unfortunately, there are currently no applicable emission
measurements available for high speed engines.
4.6.2 Temperature in NOX formation process
By looking at the variation of the emission coefficients in Figure 11, it is apparent
that reaction time alone isn‟t sufficient for accurate estimates. Kuiken suggests that the
role of ambient temperature in the reaction is so profound, that the amount of formed
particles per second approximately ten-folds if the temperature is increased by 100
degrees. This suggestion is backed up by measurements by Sarvi, who presented that
the peak temperature would be the main factor for emissions. On the other hand,
47
it was also suggested by Sarvi that the speed of the engine affects the temperature as
well.
Clearly, the engine load is affecting the amount of emissions, but most likely
through reaction time and combustion temperature and thus the modeling is
currently reasonable to be based on the IMO tier I curve alone. If more emission
factor measurements are made available, especially ones that are conducted with high
speed 4-stroke engines then a more reliable model for emissions could be
implemented based on the idea of reaction time. The effect of this implementation
would most likely result in a notable increase in the amount estimated emissions
(10-20%) near harbor areas, where ships tend to sail with low engine load. It is likely
that the combustion temperature of the engine is not something the model can or
should be able to do and thus should be left out of the emission modeling.
4.7 Alternative marine fuel – Liquid Natural Gas (LNG)
IMO and EU both seek to decrease emissions arising from marine traffic by enforcing
legislations and directives, for example, by setting a maximum limit for the amount of
sulphur in marine fuel-oil. Other example is the IMO tier III (presented in Figure 4)
curve that defines the maximum emission factor for engines installed after January
2016. Currently, there are a few main options to achieve such a drastic reduction of
approximately 80% in emissions – some involve the use of various abatement
methods presented in Chapter 2 and some include the usage of alternative fuels, such
as liquid natural gas, LNG.
The use of LNG fuel practically eliminates all of the emissions and due to lower
burning temperatures, also emissions are reduced significantly. LNG fuel does not
contain any sulfur and thus, all of the emissions are eliminated. In addition, using
LNG incorporates no risk of environmental or safety hazards (excluding from high
pressure) as it needs very specific temperature and pressure -conditions in order to be
volatile in its gaseous form. However, LNG must be maintained cold to remain a liquid
and to fulfill that requirement the LNG is to be stored at about 100°K as a boiling
cryogen, which inevitably reduces the efficiency of the overall solution. Also, the LNG
vapor boil off produced during changes of state must be let out to allow the storage
temperature to remain constant (Zbaraza, 2004). Because of this, LNG is becoming
48
increasingly popular among LNG tankers themselves, as it is possible to salvage large
amount of boil off vapor for propelling power.
4.7.1 Emission modeling with ships using LNG fuel
In Table 3, a comparison of emission factors between LNG and diesel according to
Wärtsilä is presented. The emissions are reduced approximately 85% per kWh
and also emissions are measured to be significantly lower. According to (Zbaraza,
2004) emissions are reduced approximately 70% and in a recent report from DNV
it was suggested that the reduction is almost 100% (DNV, 2011).
Table 3: and emissions from a duel fuel engine according to Wärtsilä. For
both operation modes (diesel and gas) measurements were made using two different load points.
Diesel
LNG
Engine Load 0.75 1 0.75 1
22g/kWh 11.5g/kWh 2g/kWh 1.4g/kWh
630g/kWh 630g/kWh 450g/kWh 430g/kWh
Typical LNG fuel consists mostly of methane (92-99.7%) and ethane (0.1 – 9.3%) and
therefore approximately 75% of the mass in LNG fuel is carbon. Therefore, in perfect
combustion process, in which every carbon atom is assumed to be oxidized, one gram
of LNG fuel produces approximately 2.76 grams of . Diesel fuel, on the other
hand, contains more carbon by mass (approx. 85%) and produces 3.04 grams of
per burned fuel gram. Also, LNG has a higher heat value than diesel – Comparing the
lower heat values (LHV) reveals that LNG combustion produces 14% more heat. Using
the higher heat values (HHV) that takes into account also the residual heat in the
exhaust gas, the difference between the two fuels is even more distinctive: 18% more
heat from LNG if all of the residual heat from exhaust gas gets successfully utilized
from using both of the fuels.
Currently, most of the marine engines that are capable of using LNG fuel are so-called
dual-fuel engines and are designed to use small amounts of marine diesel oil, especially
with low engine loads. Measurements made by Wärtsilä suggests that the shape of a
SFOC curve for dual fuel LNG engine is similar to diesel engines, although the fuel
consumption increases more steeply in the lower engine load reaching up to 175% with
49
very low engine loads (Zbaraza, 2004). Indeed, because of this, current dual fuel
engines prefer to run fully on diesel if engine load drops below 40%, which would make
emission modeling more complicated. The overall efficiency of the dual fuel engine is
also more than 20% lower when compared to diesel engine, but the efficiency ratio is
likely to increase and ultimately challenge diesel engines after the relatively new LNG
technology has been adopted properly.
Assuming the lower efficiency of the current LNG engine technology (-23%), the
relative emissions reduction potential of LNG compared to diesel can be
calculated to be:
(
)
(42)
(
)
where is the amount of produced by burning one gram of LNG,
is the respective amount by burning marine diesel oil and is the ratio of
heat values of the fuels. Thus, using LNG for marine vessels might result in 7.3%
reductions for as a worst case scenario. Let‟s assume the same overall efficiency for
using both fuels. Then the relative reduction is equal to
(
)
(43)
Using the higher heat values i.e. assuming that all of the residual heat can be captured
from the exhaust gas, the theoretical upper limit for reductions can be calculated to
be
(
)
(44)
Depending on calculation method and efficiency of the LNG engine, the reduction
potential for varies from to a maximum of 26.8%. Because all of these
presented beneficial aspects in using LNG fuel in marine vessels, policy makers are
currently very interested in getting their hands on scenario analysis about partial and full
adoption of LNG technology. Therefore, LNG as a marine fuel has been added to the
50
emission model STEAM2 using a rough average reduction of 20% for assuming
complete reduction in emissions. If LNG technology increases its popularity, then
it might be worthwhile to model the use of LNG fuel in a more sophisticated manner,
which might involve applying a steeper SFOC curve for LNG fuel and also taking into
account the dual-fuel engine‟s diesel fuel usage with lower engine loads.
4.8 Route interpolation with a shortest path algorithm
As a result of AIS blackouts, sometimes the travel route interpolation has to be done
over long distances using a straightforward line connecting the two locations. Inevitably,
these lines often go across continents and islands. Because of this, also the estimated
speed for this route is significantly underestimated when a constant speed estimate for
the line route is calculated. The other simple option, to use the last known speed data
for the route, might naturally result in even larger margin of error. Sometimes the cause
of this problem is not just temporary as it is with AIS shortages – there are significant
geographical gaps in AIS coverage as well. Indeed, if STEAM2 emission model would
be used in the Mediterranean Sea or the North Sea, a more sophisticated route
interpolation feature would be most welcome.
To resolve the interpolation problem, a shortest path algorithm can be used to calculate
the most probable and plausible marine route connecting any two end points in any sea
region. To save computational time, however, this alternative interpolation method is to
be used only if the result from straight line interpolation would be evaluated unsuitable,
for example, in case the line interpolation would suggest a route through land.
The main idea of the algorithm is to use a node grid that defines the sea surface – A
node has a value of 0 if it is on land and otherwise has a value of 1 indicating a marine
region. To generate the node grid, a Matlab function was produced, which evaluates a
RGB (red, green, blue) value for each pixel in a selected map-picture. In the picture,
any sea surface has been distinctively colored, for example, in pure red. The resulting
binary node grid is then repeatedly used by the model to check weather a node or a
coordinate is on the sea surface or not.
The problem of finding the shortest path is also solved using the same node grid. The
shortest path algorithm itself is beyond the scope of this thesis and the related shortest
path network optimization problem is discussed in greater detail in (Johansson, 2011).
In short, the problem is formulated as follows: There exists and outbound arc from any
51
marine node to each adjacent marine node (maximum of 8 out bound arcs including 4
diagonal arcs). Each arc is associated with a cost of traveling through it, and the cost of
doing so is set to be equal to the Euclidean distance between the two nodes. The
shortest path is generated using a Djikstra label setting shortest path algorithm which
initially associates infinite costs of travelling from the source node (s) to any other node
marking the current travel costs with labels. Then the problem is reduced into finding
routes with smaller total costs to each node, and in case a better route has been
identified the current label is rewritten to match the cost of the better route. The
process is repeated until the labeled cost to the target node can no longer be made
smaller.
Figure 12: Shortest path example from to using a 100x100 node grid. The solved
shortest path is presented with an orange line and the route interpolated with the regular method is presented as a yellow line. Map is provided by Google Earth.
In Figure 12, an example route produced by the algorithm is presented. In this
example, a ship‟s last message before an unknown blackout occurred was sent in the
place marked as (source) near Vaestervik of Sweden. After 20 hours later the ship
continues sending AIS messages from (sink) in the Gulf of Riga. Using the two
endpoints the model is able to calculate the shortest and thus a probable route from
to without traveling through land. The final cost label for the node given by the
algorithm indicated that the minimum travel distance is 472 kilometers and therefore
the ship is assumed to travel the distance with a constant speed of 6.6 m/s. The
traditional interpolation method using a straight line connecting and leads to a travel
52
distance of 7 degrees in longitude, which translates to 420 km ( in latitude) with a
constant speed of 5.8 m/s respectively.
In this example, a very rough node grid of 10 000 nodes covering the whole Baltic Sea
was used. Due to the scarcity of the grid, the resulting shortest path (orange line)
problem was solved in a fraction of a second. In contrast, solving the problem using a
more detailed grid map of 500x500 nodes would take more than a minute using a
modern PC but would produce smoother routes with better resolution. Indeed, it is
crucial to use the more intelligent interpolation feature only when it is needed. For
instance, the method could be used only if
The vessel in question has a substantial impact on emissions
The distance between and is large enough
A normal interpolation method would result in route that travels through land
Using the presented criteria it is possible to increase the resolution of the method if
needed without causing unnecessary increase in total computation time. The first
criterion could be determined by checking if the vessel has a valid IMO number or not.
A proper cut-off distance for the second criterion could be set to equal 10-50 km. The
third clause has been already programmed to the algorithm using the logical node map
by checking all the non-marine nodes, if any, in between and Finally, the
computational effort required by the algorithm can be significantly reduced by using a
set of different maps with different resolution and geographical scope.
53
5. Model evaluation
In this chapter, the extended model is evaluated against available experimental data.
Selected numerical results, including the comparison of annual fuel statistics and model
estimates are presented.
5.1 Evaluation of the predictions of STEAM2 for engine power
An example comparison between the predictions on main engine power of the two
model versions is presented in Figure 13. The engine power data has been measured in
this study at the engine room of a large RoPax (Roll On – Roll Off cargo/Passenger)
vessel. The presented voyage was done in an archipelago area near Stockholm, Sweden,
and in the vicinity of this archipelago, in April 2008. This specific dataset is used, as it
was the only one available in the Baltic Sea region. Measured power profiles, such as
the one presented in Figure 13, are difficult to obtain, as only a limited number of
vessels have internal equipment suitable to collect this data.
Figure 13: The predictions of the STEAM2 model and the corresponding measured engine power. The data has been measured for a 60 000 t RoPax vessel that was sailing in the Baltic Sea within and near the archipelago surrounding the city of Stockholm in
April 2008.
54
The basic statistical measures of this comparison are presented in Table 4. The
predicted main engine power of model is in a fairly good agreement with the measured
values. The predictions of the STEAM2 model are moderately better that those of
previous model version in terms of the mean absolute error, and vice versa in terms of
the mean error. STEAM2 slightly under estimated the engine power. There are
physical factors that have been neglected in both models, such as the influences of the
sea ice on the kinetic energy of the ship, the squat effect and the sea currents. The
model would therefore be expected to under-predict the required engine power.
Table 4: Statistical measures for the power predictions of STEAM2 model. P is the
predicted power, is the measured power and the number of observations n = 729.
Errors in percent in the table have been computed with respect to the mean values of the measurements.
Formula STEAM2 Measured (M)
Mean value
∑ 11190kW 12338kW
Mean Error
∑( ) -1148kW (-9.3%) -
Mean Absolute Error
∑( ) 1845kW (15%) -
The Hollenbach method used in STEAM2 results in a steeper power curve compared
with the corresponding method in the previous model, i.e., a relatively lower resistance
for low ship speeds and a higher one for high speeds.
As it was discussed in Chapter 4.4, the rapid changes in the unprocessed speed data
causes problems in estimating current power consumption. Indeed, as it can be seen
form Figure 13 the measured power is smooth whereas predicted power is not because
of the rapidly changing speed data. The difficulty of modeling instantaneous power and
especially changes in engine load arises from the initial problem of estimating engine
power based on speed data itself – changes in speed are interpreted as a result of
changes engine load but the reality is likely quite the opposite: engine load is usually
kept steady for economic reasons and the forces acting on the ship causes the speed to
fluctuate. Based on this study, smoothing the speed data by taking the average of the
two end point values should be an improvement in itself.
55
5.2 Evaluation of STEAM2 predictions for fuel consumption
The reported and predicted fuel consumption of a RoPax ship in 2007 has been
presented in Figure 14a-b. The extended model predicts the total fuel consumption
fairly accurately and slightly over-predicts the fuel consumption of auxiliary engines and
boilers, which is a significant improved as the older model version substantially over-
predicted (by more than 150%) the latter consumption.
Figure 14a-b: The monthly average fuel consumption of a RoPax ship in 2007, as reported by the ship owner, and predicted by the two model versions. The total fuel consumption is presented in the upper panel, and the fuel consumption of auxiliary
engines and boilers in the lower panel.
0
500 000
1 000 000
1 500 000
2 000 000
2 500 000
3 000 000
3 500 000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
Co
nsu
mp
tio
n [
kg]
Total fuel consumption
Shipowner
STEAM2
0
50 000
100 000
150 000
200 000
250 000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
Co
nsu
mp
tio
n [
kg]
Fuel consumption of auxiliary engines and boilers
Shipowner
STEAM2
56
Similar studies were made with four other RoPax-ships. The results are presented in
Figure 15. The overall accuracy of the aggregated predictions is satisfactory. As it was
discussed in Chapter 4, the effect of currents would likely reduce the fuel estimates.
The addition of kinetic energy, would suggest an increase of approximately 4% and the
effect of ice during winter would increase estimates as well.
Figure 15: The reported and predicted total fuel consumption for five RoPax vessels from January to November in 2007. The vessel RoPax 4 is the same ship, the data of
which has been presented in Figures 7a-b.
5.3 Evaluation of the modeling of load balancing in STEAM2
The model determines the number of engines, which need to be operated to overcome
the predicted resistance of the ship, and the engine load of all running engines. The
validity of this feature was tested using the measurement data from the cruise presented
in Figure 16a-d, in which there were four identical main engines in the vessel
considered. The overall accuracy of predicted engine loads is fairly good or good for
most of the time in the cases presented. However, there is some inaccuracy in the initial
stages of the voyage, and for the fourth predicted engine (i.e., the one used only for very
limited time periods). The spikes in the fourth engine‟s estimated power output suggests
that the total power requirement was on the threshold for start-up of engine 4 on several
occasions according to the model. In reality, of course, large diesel engines cannot be
shut down and start-up in such a frequent manner. In a recent study (Winnes, H.,
Fridell, E., 2010) it was observed that the startup-of marine engines increases the
0
5 000 000
10 000 000
15 000 000
20 000 000
25 000 000
30 000 000
35 000 000
RoPax1 RoPax2 RoPax3 RoPax4 RoPax5
Fuel
Co
nsu
mp
tio
n [k
g]
Ship owner
STEAM2
57
amount of emissions significantly for a short duration. If such a feature would be
included in the model later on, then it might not be meaningful to use the feature with
multi engine setups.
Figure 16a-d: Predicted and observed engine loads of four identical main engines in a
large RoPax ship. The time scale for all plots a-d is the same, presented in panel (d). MEx, x =1,2,3,4, are the four main engines. „Estimate‟ refers to the prediction of
STEAM2. The numbering of the main engines in the model has no influence on the engine load predictions; for instance, in panel (b) the curves ME2 (estimate) and ME3
(observed) are directly comparable.
58
5.4 Evaluation of the PM emission factors
The emission factor predictions by STEAM2 are compared with measurements
available from literature in Figure 17a-d. The engines loads and fuel sulphur contents in
these studies are as follows: 85 % and 2.85 % (Agrawal et al., 2008), 84 % and 1.90 %
(Moldanova et al., 2009), 85 % and 2.21 % (Petzold et al., 2008), and 57 % and 3.01 %
(Murphy et al., 2009). For simplicity, these studies are in the following referred to as
AGR, MOL, PET and MUR. The engine load is within the commonly used operation
range for the three first-mentioned studies, but it was substantially lower in Murphy et
al. (2009). The sulphur content of fuels varies from 1.9 to 3.0 %.
For a substantial fraction of these predictions, STEAM2 is in agreement with the
measurements; the agreement is best in case of AGR. However, there are also
significant differences. The most significant differences are found in comparison with
the data by MOL, especially for and . The predicted sulphate emission factor is
approximately three times larger than the measured value. According to MOL, the
measured low sulphur conversion to sulphate may be a result of the relatively smaller
amounts of V and Ni in the fuel, compared with, e.g., AGR. The catalytic properties of
Ni and V enhance the sulphur conversion to sulphate.
According to Petzold et al. (2010), the conversion efficiency of fuel sulphur to
particulate sulphate is linearly increasing with increasing engine load from 1 to 5 %
(such a dependency is not allowed for in STEAM2). This could be one of the reasons
for the deviations of predictions and data in case of MUR, due to the low engine load.
A detailed investigation of the complete data set of Petzold et al. (2010) using STEAM2
reveals an increasing difference in to particulate conversion with decreasing
engine loads.
In case of MUR and AGR, the ash emission factor was computed from the ash content
of the fuel, whereas MOL and PET report directly measured values of ash. These ash
emission factors are therefore not directly comparable with each other, and the MUR
and AGR ash emission values are strictly speaking not comparable with the STEAM2
predictions. There may be processes during fuel combustion, which lead to changes in
the amount of emitted ash. MOL reports the highest ash emissions, although the ash
content of the fuel used by MOL is the lowest. In comparison with PET, the STEAM2
ash emission factors are in a good agreement. The ash emissions in principle depend
59
on the ash content of the fuel, but this is not taken into account in the model. However,
one cannot conclude based on the above comparison of predictions and data that this
would be a significant impact.
Figure 17a-d: Comparison of the predicted and measured emission factors for the chemical constituents of PM. The measured data has been extracted from Agrawal et
al. (2008), Moldanova et al. (2009), Petzold et al. (2008) and Murphy et al. (2009).
60
The water content of in these four datasets varies significantly. This can be due to
differences in the experimental setups, sampling conditions and reporting. Water and
organic compounds may condense on particulate surfaces after fuel combustion.
Dilution and cooling of the sample to a lower concentration and temperature have
an effect on the amount of condensed water. The amount of water is commonly
calculated assuming a constant ratio of and water (Agrawal et al., 2010; Agrawal et
al., 2008; Petzold et al., 2008). To overcome these difficulties, a dry mass could be
used instead; however, this would require the inclusion of aerosol condensation
processes. In STEAM2, the associated water is modelled separately (according to the
IMO GHG2 study), and the user has an option to exclude it.
61
6. Emission analysis for the Baltic Sea shipping
In this chapter the results produced by the model for marine shipping in the Baltic Sea
are presented, which cover the time interval between January 2006 and December
2009. During these years, the maximum sulfur content of marine fuel has been forced
to a lower level, economic recession took place and the shipping traffic itself has
evolved and shifted. These geographical and quantitative changes in the shipping sector
and in their produced emissions are illustrated in this chapter.
There is an ongoing discussion in the International Maritime Organization about the
options to curb exhaust emissions of international shipping. These are mostly revolving
around the issue of climate change impact and emissions of , but the principles
behind different market based instruments also apply to other pollutants like ,
and . According to the study of ENTEC, several options with different benefits and
drawbacks are available: First is the “polluting state pays” scenario, which builds on the
idea of allocating the emissions of ship traffic according to the flag state of the ship or by
geographical area where the actual emissions occur. With these options however, the
individual states would still need to find a way to allocate the costs to individual ships
eventually, which could be done by imposing taxes using ship specifications such as
weight taking travel distances into account. This in turn would require accurate
information about how these ship attributes affect the emissions of individual ships.
Another option is an emission trading system (ETS), which relies on emission credits, a
system much like to the existing credit mechanism already applied in Europe, but
in which the shipping sector is not currently included. A more direct option is based on
the actual fuel consumption of ships, in which the additional cost is built in to the
bunker fuel prices. To illustrate the implications of implementing some these cost
allocation methods, the emission shares have been evaluated in this chapter using
several of the above mentioned criteria.
6.1 Emissions from Baltic Sea shipping in 2006-2009
The archived AIS data for the years 2006 to 2009 presented in Chapter 4.4 was used to
estimate monthly and annual emission estimates for the Baltic Sea. It was observed that
the ships encountered in the AIS data can be classified based on their fuel consumption
and IMO number. The monthly appearances of these classes are presented in Figure
18.
62
Figure 18: Monthly number of ships according to the archived AIS data in 2006 and 2009. Active ships with IMO number represent the regular and registered marine traffic while ships without the IMO number represent unregistered vessels and are not found
in the internal ship database. Non-active ships is the group of vessels appearing in the archived AIS data which do not consume any fuel according to the model or send more
than one AIS message in a month.
Ships with an IMO number represent the commercial and regular ship traffic while the
rest of the ships are not specified in the internal ship database and are assumed to be
small vessels and are thus attributed with generic small tug boat specifications. Besides
the data from these active ships, the archived AIS data contains singular messages or a
small amount of physically impossible route data from some vessels. Almost all of these
“non-active” ships showing in the archived AIS data are of unknown origin but seem to
appear more frequently during summer months. In 2009 also the number of active
ships without IMO number shows a seasonal dependency which describes the
increased passenger and yacht traffic during the summer period in the Baltic Sea area
(Figure 18). Moreover, it can also be seen from the figure that while the regular and
registered ship amount on the whole remains unaffected by season the total number of
vessels are at maximum during summer months due to the increase in non- commercial
traffic.
The marine traffic was affected by the economic recession starting from 2008 and
continuing throughout 2009. Most of the riparian states showed decreasing gross
domestic product (GDP) rates starting from the second quarter of 2008 with a few
exceptions: for Denmark the economic difficulties began in the second quarter of 2006
63
and in Russia after a steady season of growth the inflation adjusted annual GDP growth
rate crashed quickly below -9% in the second quarter of 2009 (Trading economics,
2011).
The annual emission estimates according to the model for 2006 – 2009 are listed in
Table 5 and monthly time series for , , and emissions covering the
whole study period are presented in Figure 19.
Table 5: Emission estimates for 2006 – 2009 in the Baltic Sea. AIS downtime has been taken into account in the figures. Estimates for 2006 are presented in tons and the
following years in relation to emissions in 2006. PM emissions are assumed to be equal
to the sum of OC, EC, Ash particles and with its associated water molecules.
Pollutant
2006
[tons]
2007
(of 2006)
2008
(of 2006)
2009
(of 2006)
Gases 16 490 000 110.50 % 113.90 % 108.90 %
335 900 109.90 % 112.30 % 107.10 %
144 200 91.30 % 91.40 % 86.20 %
51 640 112.60 % 124.90 % 124.50 %
PM constituents 5 730 109.90 % 112.60 % 107.50 %
2 220 109.90 % 113.00 % 108.10 %
1 620 109.80 % 112.80 % 107.70 %
20 930 91.40 % 91.90 % 86.20 %
PM
30 500 97.20 % 98.50 % 92.90 %
In 2006 a total of of was produced by the marine shipping in the
Baltic Sea according to Table 5 while only 5% of the total emissions resulted from
the ship traffic without certified IMO number. The year 2007 showed a steady increase
in overall marine activity resulting in a 10% increase in emissions. To put this
amount of produced into perspective however, the contribution of the traffic in
Baltic Sea to the global emissions can be estimated to be less than 2 percent
according to the second IMO greenhouse gas study in which the total emissions
were estimated to total in 2007 (Buhaug, 2009).
64
Figure 19: Monthly estimates for , , and PM emissions during 2006-2009. PM and
CO emissions have been multiplied by 5 for enhanced visual clarity.
The overall trend of , and emissions was increasing from 2006 to 2008,
but decreased during 2009 in relation to the respective emissions in 2008 because of
the recession. Regular passenger traffic was least affected by the recession, whereas all
of the emissions from containerships and vehicle carriers showed significant decrease
signaling a slowdown of business in these sectors. The decrease of all the modeled
emissions during 2008 - 2009 was nevertheless no more than 5 % despite the significant
decrease of all modelled emissions from Roll-on/Roll-off ships (RoRo), general cargo
and containerships; with these particular ship types the modelled emission amounts
decreased up to 20% and no less than 14%.
At the end of 2009 a total of of was produced. Due to the increase
in the number of unspecified ships during the past years the share of IMO certified
ships of the produced emissions decreased to 90%. Moreover, the strong
simultaneous increase in emissions can be explained with the large number of these
new unspecified ships; small boats produce relatively large amounts of according to
the model.
The annual emission amounts of and in respect to the reference year of 2006
have evolved differently than the previously discussed and because of the
change in SECA fuel sulphur requirements: in May 19th
of 2006 the maximum allowed
65
sulfur content used in SECA area was decreased from 2.7% to 1.5%. The effect of this
reduction is clearly visible in Figure 19. Also a clear seasonal trend in the presented
emissions for every year of study can be identified from the figure – for example, in any
year of study the (sulfur independent) emissions in July are 16% - 25% larger than in
January. The main reason for this is not just the seasonal increase in ship amounts
without IMO number (Figure 18) but rather the increase in fuel consumption of the
more notable IMO certified ships; e.g. Roll-in/Roll-out passenger ships (RoPax) which
are later shown to consume more than one fourth of the total fuel consumption, seem
to be significantly more active in the summer time than in winter. Most of the ship types
besides RoPax and passenger ships, however, do not show a strong seasonal
dependency in their activities.
6.2 Geographical emission changes from 2006 to 2009
Besides the total emission estimates, the STEAM2 model is also capable of presenting
the cumulative geographical distribution of the modeled emissions (Jalkanen et al.
2009, 2011). During 2006 – 2009 the annual fuel consumption of marine traffic in the
Baltic Sea has increased approximately 9%. At the same time, the reduction of
maximum sulfur content of marine fuel has caused reductions in the total and
emissions.
To be able to identify where these changes have affected the local air quality the most, a
difference map using the aggregate emission data from January - April of 2006 and
2009 was produced (Figure 20a-b).Comparison of the combined monthly emission
estimates between January and April in 2006 and 2009 reveals that emissions in
2009 are 37 % smaller than in 2006. A similar comparison for emissions shows a
decrease of 26% respectively. However, it can be seen from the figure that the majority
of these reductions reside in the major ship travel routes away from dense human
population.
Moreover, as the several areas with positive emission amounts indicate, emissions
in some harbors, most notably in Gdansk and Kiel, have even increased during the
study period. One explanation for this result is that the use of auxiliary fuel
consumption (consisting of 0.5% sulfur in mass according to model) which is
concentrated near harbor areas, is unaffected by the imposed sulfur reducing
legislation.
66
Figure 20a-b: Estimated change in (a) and (b) emissions distribution between
2006 and 2009. Total emissions of January - April 2006 have been subtracted from the estimated total emissions of January -April 2009. Green colors indicate the areas where estimated emissions in 2006 surpass the emissions of 2009, while other colors indicate a increase in emissions. The color scale corresponds to emissions in kilograms in an area
of 0.03 x 0.03 degrees (approximately 6.5 ).
67
A corresponding difference map was also produced using emissions (Figure 20b).
In addition for evaluating changes in emission distribution during the study
period, it also possible to indirectly identify geographical changes in fuel consumption
and travel routes from the figure. The reason for this possibility is that the aggregate
emissions correlate very strongly with aggregate (and even instantaneous) fuel
consumption in the model. Unfortunately no emission grid has been produced for
emissions, which would‟ve been the ideal choice for identifying geographical changes in
fuel consumption.
A couple of significant new ship routes (e.g. near Helsinki and Tallinn) can be
identified in areas where emissions in 2009 surpass the estimated emissions in
2006 considerably. Also, many harbors near Denmark, Germany and Stockholm area
in 2009 seem to support more marine traffic than in 2006 which might be caused by the
increase in small and unidentified ships that most likely operate near the coastline.
Furthermore, the marine traffic seems to have been increasing near the Baltic States, St.
Petersburg and especially near Helsinki which is partly because of the opening of the
new Vuosaari harbor for goods traffic in 2008. The area between Gotland and Oland of
Sweden and the whole Gulf of Bothnia are the most notable geographical areas where
emissions and thus presumably the marine traffic has decreased during the study
period.
6.2.1 Allocation of emission costs by geographical area
One of the suggested options for emission allocation is based on geographical
distribution of emissions, in which the emissions happening inside the economic zone
of each country would be included in the national reporting of emissions. Confirming
the actual level of emissions with measurement inside each economic zone, however,
would be challenging as there are at any given time over 2000 ships sailing the Baltic
Sea. Confirming the emission levels through measurements will cost a lot of money and
personnel resources and still leaves doubts how to allocate emission to each country.
Fortunately, emission distribution estimates by geographical areas such as in Figure 20a-
b can be produced with models based on AIS messages such as STEAM2.
6.3 Flag state analysis
The MMSI code contains information about the flag state of the ship thus enabling the
comparison of nations in terms of produced emissions of their marine traffic. In Figure
68
21a, based on the total fuel consumption eleven most contributing flag states for each
year in study are presented.
Figure 21a-d: Estimated total fuel consumption (a), CO emissions (b), payload (c) and the fleet size (d) of the eleven most contributing flag states according to flag state
information of the MMSI codes. Due to variable AIS data quality and non-optimal travel distance calculation method used in 2006, the respective transferred payload
values may have been overestimated up to 20%.
The changes in fuel consumption of the riparian states are similar to their respective
changes in GDP with few exceptions: Sweden, which is the undisputed leader in total
fuel consumption, is showing steady increase even in 2009 despite the recession. Also
Denmark‟s growth in terms of fuel consumption has been steady and rapid despite its
nonexistent GDP development at that time (Trading Economics, 2011). The fleet
sailing under the flag of Russia at the Baltic Sea is suspiciously small, resulting to
emission estimates that are comparable to those for Estonia. Moreover, the total fuel
69
consumption of the Russian fleet has decreased during the years in study which
coincided with a strong growth period in the Russian economy. Interestingly the ship
traffic near Russian harbors presented in Figure 21b is not showing the same decreasing
trend.
Aggregate , , and emissions of fleets tends to correlate strongly with it‟s
aggregate fuel consumption and thus the relations of these emission amounts between
the flag states would be similar to those of in Figure 4a and are not presented in this
paper. emissions however do not share this similarity which can be seen from
Figure 21b: If measured in emissions, Denmark is the second largest polluter in the
Baltic Sea and Denmark‟s and Sweden‟s large share of indicates that a large portion
of the unidentified small vessels are from Denmark and Sweden. Indeed, the Riparian
states and especially Sweden, Germany and Denmark account for most of the new
ships appearing in the AIS data which can be seen from Figure 21d.
The ships sailing under the flag of Bahamas, Cyprus and Malta are heavy compared to
other states: the average gross tonnage (GT) for a Maltan ship is close to 11200 tons
and for Bahamas over 14000 tons whereas the average weight for a Swedish ship is less
than 2800 tons (in 2009). Furthermore, the abovementioned flag states are also more
cargo oriented; From Figure 21c it can be seen that a very large portion of the
transferred payload is carried by the fleets of Malta, Bahamas and Cyprus alone. These
payload values have been calculated using a ship type specific fraction of deadweight,
given by the Technical research center of Finland (VTT, 2011).
The average age of the fleet and its recent development differ among the most
contributing flag states significantly. For example, the Russian fleet is the oldest on
average in 2009 (25 years) and more than 14 years older than the youngest fleet of
Netherlands although the fleet of Finland and Sweden is almost as old as Russia‟s.
6.3.1 Allocation of emission costs by region or flag state
It is likely that allocating the emissions to the flag state would eventually lead to a
situation where the ships in the Baltic Sea will be reflagged to a different country with
less strict policies. This would happen even more likely if the flag state emission
allocation is done only on regional or European level. The flag state allocation would
probably only work if done simultaneously globally in a way that the same rules apply to
all flags.
70
6.4 Emission analysis by ship type and size
According to the model results, the marine traffic in the Baltic Sea is for the most part
dominated by eight most contributing ship type classes present in the model‟s ship
database; The top eight accounts for 93% of total emissions in all classes and fuel
consumption which applies for all the years in study 2006 - 2009. These ship types and
their average attributes in 2009 are presented in Table 6.
Table 6: Most common ship types among the Baltic Sea marine traffic and their average attributes in 2009 according to model shipping statistics. Cargo payload of deadweight tonnage (DWT) is the average fraction of deadweight that is allocable to payload based
on calculations by VTT. For containers and RoPax ships, the unit emission is
dependent on GT. Unit emission is the estimated amount of emissions per
transferred payload.
2009 ROPAX TANKER GC CONTAINER RORO BULK PASSENGER
Average GT
[ton] 16 560 27 380 4 680 20 770 15 010 25 800 18 440
Average DWT
[ton] 3 290 47 140 6 390 24 060 9 030 44 600 2 110
Payload of DWT 0.42 0.5 0.4 0.4-0.65 0.24 0.5-0.6 -
unit emission
[ ] 127 8.51 30.6 26.0 67.7 7.32 -
Average age 19.7 8.6 15.8 8.5 15.5 13.9 30.4
Avg. main engine
power [kW]
14 700
8 310
2 730
15 660
10 780
7 710
12 440
Avg. service speed
[m/s]
9.1
6.7
6.3
9.8
8.8
7.2
7.7
Common engine
design 4-stroke 2-Stroke 4-stroke Both 4-stroke 2-stroke 4-stroke
Total ships in 2009
(change from 2006)
220
(-15)
1 785
(+316)
2 281
(-68)
347
(+106)
152
(-19)
984
(-100)
194
(+33)
Main purpose
Vehicle and
passenger
travel,
cruising
Liquid
cargo
transfer
General
cargo
transfer
Container
cargo
transfer
Vehicle
transfer
Bulk
cargo
transfer
Cruising and
passenger
travel
RoRo and RoPax ships, besides having several hundred passengers onboard, also
transfer cargo on wheels such as trucks and cars whereas passenger ships are more
leisure oriented and packed with up to several thousand passengers. Typically the
71
payload of these vessels is small compared against the cargo ships. RoPax and
Passenger ships and ships are the most frequent and regular types which are, on
average, traveling in the Baltic Sea on 10 separate months per year. In contrast,
unspecified ships without IMO number and bulk carriers travel infrequently and these
ships appear only in 3 separate months on average.
During the recent years the order of these ship types by contribution to emissions has
been constantly changing with the exception of RoPax, Tankers, and general cargo
(GC), which have been the top 3 polluters in the order as presented for every year of
study. These emission shares in 2006 and 2009 for each ship type are presented In
Figure 22a-b.
Figure 22a-b: Share of emissions travel and payload in 2006 (a) and 2009 (b) of the
most contributing ship types. , PM and emission shares according to model
are approximately equal and differs at most unit percent from the presented value.
S/U refers to small tugs and unspecified ships.
72
From Figure 22a-b it can be seen that the heavy cargo ship classes are not to be blamed
alone to have caused most of the emissions. Indeed, bulk ships which represent the
heaviest vessels sailing in the Baltic Sea contribute approximately 5% to total emissions
of each modeled type while RoPax ships contribute as much as 25%. Furthermore, the
estimated total payload for bulk ships is four times larger than for RoPax ships in 2009.
Tankers, which are a mix of several subclasses that include chemical, LNG, liquid
petroleum gas (LPG) and other liquid product tankers, are responsible for a half of the
total payload transferred across the ocean in 2006 - 2009. Bulk carriers transfer the
second biggest amount of payload (approx. 18%), which usually consists of unpackaged
cargo such as coal or wood.
Large weight makes a ship sail deeper causing the amount of displaced water and water
surface to increase which in turn contributes to the resistance in moving in water. To
study the contribution on total emissions of the heaviest ships sailing at the Baltic Sea
the estimated emission shares were allocated among different weight classes in 2006 and
2009. The result of this study indicates that ships weighting more than 25000 tons are
responsible for approximately 27% of total , , emissions in 2006 and in
2009 approximately 33%. Indeed, the role of large ships in the production of emissions
has increased throughout the study period. For example in 2006 half of the
emissions was produced by ships weighting more than 12 250 tons in 2009 by ships
weighting more than 14 300 tons respectively. Furthermore, the largest ships (GT >
50000t) are much younger than the smaller ones being only 7 years old in the average
which further reinforces the observation of the shift towards larger ships in the Baltic
Sea. As it was discussed in Chapter 6.3 it is likely that these heavy vessels are being
registered to sail under a surrogate flag such as Malta.
In 2009 the majority of emissions (61.7%) are being produced by 2971 ships that
weight more than 10000 tons. Most of these are bulk carriers and tankers which prefer
2-stroke engines as these ships are heavy and require large power outputs. 2-stroke
engines are designed to run with relatively small RPM and because of the longer
reaction time for formation process they also produce significantly more
emissions per produced engine power output. This is the reason why bulk carriers and
tankers have relatively large shares when compared against other emissions classes
they have produced.
73
Interestingly, the usage of main engine fuel and auxiliary fuel is strongly differentiated
among the weight classes. For example, in 2009 the largest weight class (GT > 50000
ton) is responsible for using 12.5% of main engine fuel but only 4.9% auxiliary fuel.
This relation between the consumption of main and auxiliary fuel is reversed in the
smaller weight classes – The contribution of the smallest class is only 5.4% for main fuel
but 26.5% for auxiliary fuel. These relations apply qualitatively for every year in study as
well.
Even though the weight significantly affects the fuel consumption of any marine vessel,
an increase in weight is outweighed by an increase in velocity which is one of the
reasons for RoPax ships, with their high average service speed of 9.1 m/s, are having the
biggest share of total emissions. On the other hand bulk ships and tankers travel with
speeds lower than 14 and with this combination of relative low service speed and large
cargo capacity they are the most economical classes having by far the smallest unit
emission per transferred payload (Table 6).
6.4.1 Allocation of emission costs by individual fuel consumption
The option of introducing an additional environmental tax to the ship fuel prices is
probably the simplest way of allocating the environmental tax burden in the ship. This
option however poses a problem as it leaves several important administrative questions
open, like who will administer the emission fund which is collected as a part of the fuel
price.
If this method for cost allocation would be implemented, it is apparent that RoPax ships
would suffer the most and might even force some significant RoPax/RoRo dependent
enterprises out of business because of the inevitable increase in fuel prices. Also, some
of the current cargo flow over the Baltic Sea might transported by alternative methods
(e.g. by road or by train) in the future – A scenario which should be studied thoroughly.
74
7. Conclusions
In this thesis, an extended and improved model version of the original marine shipping
model (Jalkanen et al., 2009) was introduced. It‟s capability to evaluate power
requirements and fuel consumption was validated and the model was used to produce
various emission estimations and shipping statistics for the past few years. Also, further
improvements for the model were presented and difficulties arising from using AIS data
for emission estimation the model were discussed.
7.1 Conclusions from the emission estimation model
The use of the AIS data facilitates an accurate mapping of the ship traffic, including the
detailed instantaneous location and speed and of each vessel in the considered area.
The presented model allows for the influences of a comprehensive range of relevant
factors, including accurate travel routes and ship speed, engine load, fuel sulphur
content, multiengine setups, abatement methods and waves. The presented model is
the only method in the available literature that includes such a range of effects.
In previous emission inventories of marine traffic, constant emission factors have
commonly been used. However, in order to obtain accurate predictions, at least the
dependence of shipping emissions on engine load has to be taken into account. This is
especially important in port areas, as the European sulfur directive (EC/2005/33) states
that the fuel used in EU harbor areas must not contain more than 0.1 % sulfur since the
beginning of 2010. This directive will have a significant impact on the emissions
from ships at berth, which should be taken into account by any model used in local
scale modeling of harbor regions. It is important to be able to reliably evaluate the
effects of the policy options that focus on reducing the emissions from ships. The
health and climatic influences can be substantially different for the various chemical
constituents of ; the modeling should therefore disaggregate the emissions from
ships accordingly.
The relatively largest uncertainties of the model predictions presented probably arise
from the use of various types of fuel (Hulskotte and Denier van der Gon, 2010). It is
challenging to extract the detailed data regarding the fuel types used in ships in various
geographical areas. However, if the data is available on the fuel type or the sulphur
content, the model can adjust itself accordingly, and provide emissions, facilitating also
various abatement strategies. Another challenge is the scarcity of detailed composition-
75
resolved experimental data on emissions. The emissions of the chemical
components of should be analyzed at various engine loads, and using various fuels,
in order to be able to more comprehensively analyze and evaluate the performance of
the modeling approaches. Finally, in the future the model should be properly validated
against direct emission measurements and not just instantaneous power measurements.
However, some direct emissions measurements taken above a marine vessel have
already been analyzed and the model‟s prediction accuracy compared to these data
looks promising.
7.2 Conclusions from the emission estimates for 2006 - 2009
The estimated emissions of the reference year of 2006 were presented in Chapter 6.1.
In 2007, based on fuel consumption and most of the modeled emission classes, Baltic
marine traffic increased approximately 9% in respect to 2006. The late economic
recession is clearly visible in the emission estimates – The increasing trend of Baltic Sea
shipping significantly slowed down in 2008, which resulted in just 3% more emissions
than in 2007. The recession began to affect marine traffic the hardest in 2009
diminishing total fuel consumption and emissions to a level, which was a couple of
percent lower than in 2007.
Maximum allowed fuel sulphur content was decreased in May 2006 from 2.7% to 1.5%
and the effect of this reduction was studied. Total emissions in 2006 and 2009
between January to April was compared, and even though fuel consumption was 9%
larger during the interval of 2009, emissions were calculated to be approximately
37 % smaller in that time. A similar comparison revealed that emissions had
dropped 26% respectively. However, the majority of these reductions reside in the
major ship travel routes away from dense human population. It would be worthwhile to
investigate if this reduction in sulfur dependent emissions has had a significant effect on
the health of the coastal population, and also, what has been the indirect cost of this
directive.
Furthermore, in the forthcoming years the sulfur content of the marine fuel is to be
further reduced, possibly to a final level of 0.1% in the SECA area. It has been
estimated that the resulting increase in fuel prices after such a reduction will induce
costs of several billion euros for the Baltic shipping. In Chapter 6 it was discussed that
there are several methods how the costs of this directive can be allocated - With quite
76
different outcomes. Therefore, it would be very important to estimate the health
benefits and costs of the possible changes due to the directive thoroughly.
Analysis by flag state showed a steady increase in terms of fuel consumption from
Sweden despite the recession, in contrast to Finland and Germany. Denmark is closing
in on Finland in produced emissions. Indeed, Denmark‟s growth has been rapid
and if this trend continues unchanged, Denmark should surpass Finland in the
forthcoming years in all produced emission classes. In contrast, the fleet sailing under
the flag of Russia is surprisingly small which results to emission estimates that are
comparable to those for Estonia. Moreover, the Russian ship traffic has been steadily
decreasing according to AIS data while the geographical emission distribution near
Russian harbors does not support this downshift in Russian marine activity in the Baltic
Sea.
The ships sailing under the flag of Bahamas, Cyprus and Malta are very heavy and
cargo oriented. Therefore, their share of transferred payload is significant. It is to be
suspected that increasingly more heavy vessels are switching the flag state to those
mentioned above to be able to use cheaper fuel in the expense of emissions. It might
prove worthwhile to investigate the amount of Russian marine traffic that sails under a
proxy flag state and also, the motivation behind this phenomenon – if emissions are to
be reduced by setting stringent legislations then those legislations should not be avoided
by changing the flag state.
Eight most contributing ship type classes account for 93% of total emission and fuel
consumption. RoPax-ships are the undisputed number one in this measurement,
followed by tankers and general cargo ships. The recession affected container ships,
general cargo and RoRo the most. Surprisingly, RoPax ships contribute approximately
17% to total transferred tonnage over the sea surface, but manages to contribute to total
emissions more than 27%. In contrast, with energy efficient tankers and bulk cargo
ships, this relationship between transferred tonnage and emissions is reversed. One
explanation for this is that RoPax ships sail with relatively high speeds and thus the
resistance from water is significantly greater for RoPax ships.
The average main engine power and weight is slightly on the increase (tankers + 10%)
while the total travel distance is decreasing. This trend indicates that ship owners prefer
bigger ships and shorter travel amounts.
77
2-stroke engines dominate the largest weight classes starting from 10000t and up and
ships in these weight classes are responsible for the majority of emissions (61.7%).
Therefore, an effective way to reduce emissions would be to reduce emissions
from these ships.
7.3 Further improvements
The extended model presented in this thesis can be further improved in several ways.
Most importantly, with slight modifications it can be used for other marine regions
besides the Baltic Sea, if proper input data for the model will be available. However, the
AIS data cannot be received across extensive sea areas, unless a satellite-based AIS
reception is used. International cooperation between maritime authorities is therefore
needed to be able to extend the model into a global scale. In case the model is
extended for other marine regions still counting on the VHF-based AIS messaging, the
implementation if more intelligent interpolation feature is needed.
The power estimation process is arguably the most important part in the model and it
has been totally renewed since the previous version; without accurate power estimates,
even the engine load modeling loses its purpose. Fortunately, the detailed Hollenbach
method‟s prediction accuracy was shown to be great. If the initial information about the
ship‟s spatial attributes are not precise however, then the Hollenbach method should
not be able to produce better results than a much more simple estimation process
would. Because of this, rather than making efforts to enhance the method for power
estimation, it might prove fruitful to model some of the more important factors that are
currently not accounted for, for example, the effect of ice and sea currents. It has been
calculated that adding sea currents would affect the power estimate even more than the
effect of waves which has been included in the model in an early phase. Furthermore,
implementing sea currents should be relatively straightforward although tides and
seasonal changes might cause difficulties at open sea areas.
The most important area for geographical emission estimates are the most densely
populated harbor areas. To be able to accurately account for the emissions near harbor
area then acceleration and kinetic energy of the ship can be taken into account. This
implementation requires the speed data to be smoothed, which in turn might reduce
the undesired variance in power estimates which are caused by even the slightest speed
changes while the ship is travelling near its service speed. Moreover, a proper use of
78
acceleration data allows the monitoring of changes in engine load and thus
emissions spikes can be accurately modeled and might even enable the assessment of
unidentified ship‟s attributes.
The chemistry involving emissions in a diesel combustion process is complex to say the
least. Furthermore, engines vary significantly in size, speed and power and thus general
rules about emission factors are difficult to establish. Still, the SFOC concept as general
driver for emissions is intuitive and the results are backed by other studies in literature.
However, the logic behind the effect of fuel consumption to emissions is currently in
slight contradiction as was presented in Chapter 4.5 and the logic behind it should be
generalized if possible. To achieve this, more measurement data for emissions
against instantaneous fuel consumption is needed, especially for 4-stroke engines. At
the same time, emission measurements from a 4-stroke engine might lead to a
more sophisticated emission modeling process based on combustion time.
79
References
[1] Agrawal H., et al. 2010. Emissions from main propulsion engine on container ship
at sea. Geophys. Res., 115 (2010) D23205.
[2] Agrawal, H., et al. 2008. Emission measurements from a crude oil tanker at sea.
Environ. Sci. Tech., 42 (2008) 7098.
[3] Andreae, M. O., Gelencsér, A. 2006. Black carbon or brown carbon? The nature of
light-absorbing carbonaceous aerosols. Atmos. Chem. Phys. 6:3131-3148
[4] Buhaug, Ø., et al. 2009. Second IMO GHG study. International Maritime
Organization (IMO) London, UK.
[6] Cooper, D. A. 2003. Exhaust emissions from ships at berth. Atm. Env. 37 3817.
[7] Cooper, DA. 2001. Exhaust emissions from high-speed passenger ferries. Atm.
Env., 35 (2001) 4189.
[8] Corbett, J.J., et al. 2007. Mortality from ship emissions: a global Assessment. Env.
Sci. Tech., 41 (2007) 8512.
[9] Corbett, J.J., et al. 2010. Arctic shipping emissions inventories and future scenarios,
Atmos. Chem. Phys., 10 (2010) 9689.
[10] De Meyer, P., Maes, F., Volckaert, A. 2008. Emissions from international shipping
in the Belgian part of the North Sea and the Belgian seaports, Atm. Env., 42 (2008)
196.
[11] DNV. 2011. Greener shipping in North America.
http://www.dnv.com/resources/reports/greener_shipping_north_america.asp.
Visited 10.4.2011
[12] European Environment Agency. 2009. Air pollutant emission inventory guidebook
2009, Technical guidance to prepare national emission inventories, EEA Copenhagen.
[13] Eyring, V., et al. 2010. Transport impacts on atmosphere and climate: shipping.
Atm. Env., 44 (2010) 4735.
[14] Fridell, E, Steen, E, Peterson, K. 2008. Primary particles in ship emissions, Atm.
Env., 42 (2008) 1160.
[15] Harati-Mokhtari, A., et al. 2007. Automatic Identification System (AIS): Data
reliability and human error implications. The journal of navigation. 60. 373-389. The
Royal institute of navigation.
[16] Hollenbach, K.U. 1998. Estimating resistance and propulsion for single-screw and
twin screw ships, Ship Technology Research, 45/2.
[17] Holtrop, J, Mennen G.G. 1982. An approximate power prediction method.
International Shipbuilding Progress, p.166 and 253.
80
[18] Hulskotte, J.H.J., Denier van der Gon, H. 2010. Fuel consumption and associated
emissions from seagoing ships at berth derived from an on-board survey. Atm. Env., 44
(2010) 1229.
[19] Hänninen, O., et al. 2011. European Perspectives on Environmental Burden of
Disease - Estimates for Nine Stressors in Six European Countries. National Institute for
Health and Welfare (THL)
[20] International Maritime Organization (IMO). 1998. Regulations for the prevention
of air pollution from ships and NOx technical code, Annex VI of the MARPOL
convention 73/78, London
[21] Jalkanen, J.-P., et al. 2009. Modelling system for the exhaust emissions of marine
traffic and its application in the Baltic Sea area. Atmos. Chem. Phys., 9 (2009) 9209.
[22] Johansson, L. O. 2011. Reactive vehicle routing using a shortest path network
algorithm. Aalto University. Laboratory of systems analysis. Espoo
[23] Kasper, A., et al. 2007. Particulate emissions from a low-speed marine diesel
engine, Aerosol Sci. Tech., 41 (2007) 24.
[24] Lack, D. A., et al. 2009. Particulate emissions from commercial shipping:
Chemical, physical, and optical properties. Geophys. Res., 114 (2009) D00F04.
[25] Matthias, V., et al. 2010. The Contribution of Ship Emissions to Air Pollution in
the North Sea Regions, Env. Poll., 158 (2010) 2241.
[26] Matulja, D, Dejhalla, R. 2007. A comparison of a ship hull resistance determined
by different methods, Eng. Rev., 27-2 (2007) 13-24
[27] Moldanová, J., et al. 2009. Characterization of particulate matter and gaseous
emissions from a large ship diesel engine, Atm. Env., 43 (2009) 2632.
[28] Murphy, S.M., et al. 2009. Comprehensive simultaneous shipboard and airborne
characterization of exhaust from a modern container ship at sea, Environ. Sci. Tech.,
43 (2009) 4626
[29] Paxian A., et al. 2010. Present-Day and Future Global Bottom-Up Ship Emission
Inventories Including Polar Routes, Environ. Sci. Tech., 44 (2010) 1333.
[30] Petzold, A., et al. 2008. Experimental studies on particle matter emissions from
cruising ship, their characteristic properties, transformation and atmospheric lifetime in
the marine boundary layer, Atmos. Chem. Phys., 8 (2008) 2387.
[31] Petzold, A., et al. 2010. Physical properties, chemical composition and cloud
forming potential of particulate emissions from a marine diesel engine at various load
conditions, Env. Sci. Tech., 44 (2010) 3800-3805.
[32] Sarvi, A., Fogelholm, C.-J., Zevenhoven, R. 2008. Emissions from large-scale
medium-speed diesel engines: 1. Influence of engine operation mode and turbocharger,
Fuel Proc. Tech., 89 (2008) 510.
81
[33] Sarvi, A., Fogelholm, C.-J., Zevenhoven, R. 2008. Emissions from large-scale
medium-speed diesel engines: 2. Influence of fuel type and operating mode, Fuel Proc.
Tech., 89 (2008) 520.
[34] Schrooten, L., et al. 2009. Emissions of maritime transport: A European reference
system, Sci. Tot. Env., 408 (2009) 318.
[35] Schreier, M. et al. 2006. Impact of ship emissions on microphysical, optical and
radiative properties of marine stratus: A case study. Atmos. Chem. Phys., 6, 4925–
4942.
[36] Townsin, R.L. 1979. The Naval Architect, RINA
[37] Trading economics. 2011. http://www.tradingeconomics.com. USA, New York
City. Visited 21.7.2011
[38] VTT. Unit emissions of vehicles in Finland.
http://www.lipasto.vtt.fi/yksikkopaastot/tavaraliikennee/vesiliikennee/tavara_vesie.htm.
Visited in 10.7.2011
[39] Wärtsilä 46th Project guide. 2007. http://www.wartsila.com/en/engines/medium-
speed-engines/Wartsila46. Visited in 1.2.2011
[40] Watson, D. 1998. Practical Ship Design, Elsevier, Oxford, UK, pp 219.
[41] Watson, D. et al. 1976. Some ship design methods. Royal Institute of Naval
Architects. RINA
[42] Wild Y. 2009. Refrigerated containers and CA technology. Container Handbook,
Vol 3, Gesamtverband der Deutschen Versicherungswirtschaft e.V., Berlin.
[43] Winnes, H., Fridell E. 2010. Particle emissions from ships: Dependence on fuel
type, Journal of Air & Waste Management Association, 59 (2010) 1391.
[44] Winnes, H., Fridell, E. 2010. Emissions of NOx and particles from Maneuvering
Ships. Transportation Research D, 15 (2010) 204.
[45] Prockop LD, Chichkova RI. 2007. Carbon monoxide intoxication: an updated
review. Journal of the Neurological Sciences 262 (1-2): 122–130)
[46] Zbaraza, D. 2004. Natural gas use for on-sea transport. University of Science and
Technology, Cracow, Poland.
[47] Brunekreef, B., Holgate, S. 2002. Air pollution and Health. THE LANCET .Vol
360 October 19, 2002
[48] Starcrest Consulting Group. 2008. Port of Los Angeles - Inventory of Air
Emissions 2007. Poulsbo, WA
